Knowledge data fusion method based on artificial intelligence technology mixing

Abstract

The invention discloses a knowledge data fusion method based on artificial intelligence technology mixing, and relates to the technical field of data processing, and the method comprises the steps: obtaining complete knowledge data; classifying the complete knowledge data to form a data fusion set; obtaining target affiliated knowledge data; calculating an innovation coefficient of the target affiliated knowledge data to obtain a target affiliated knowledge data sequence; and fusing the target affiliated knowledge data and the subject knowledge data in the data fusion set based on the main fusion node and the auxiliary fusion node. The complete knowledge architecture is obtained by stripping the complete logic architecture in the to-be-fused data, classifying the complete knowledge data and fusing the target affiliated knowledge data and the subject knowledge data in the data fusion set, and the subject knowledge data is used as the subject architecture for subsequent fusion, so that the fusion is more ordered, and the fusion efficiency is improved. And the contradictory situation is processed, so that the final fusion effect is more reasonable.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and specifically relates to a knowledge data fusion method based on a mixture of artificial intelligence technologies. Background Art

[0002] Different platforms or different business fields each have their own data. With the development of data management and data construction, it is hoped that the data of multiple platforms and multiple business fields can be fused and connected. A knowledge graph is a structured data expression method that can efficiently present the knowledge information contained in the data. If the knowledge connection of multiple platforms and multiple business fields is realized through a knowledge graph, the efficiency of data fusion can be effectively improved, bringing improvements in business effects and computing efficiency.

[0003] However, since the types of knowledge data are uncertain, and their arrangements are irregular, and there are uncertain situations such as interlacing among them, it is difficult to fuse the data, and the fusion effect is difficult to meet the requirements. Summary of the Invention

[0004] To solve the above technical problems, a knowledge data fusion method based on a mixture of artificial intelligence technologies is provided, and the technical solution solves the problems proposed in the above background art.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows: A knowledge data fusion method based on a mixture of artificial intelligence technologies, comprising: Summarize the knowledge data to be fused into data to be fused, and strip the complete logical structure in the data to be fused to obtain at least one complete knowledge data; Classify the complete knowledge data to form at least one data fusion set, and the data fusion set is composed of main body knowledge data and at least one affiliated knowledge data; Screen the affiliated knowledge data in the data fusion set to obtain at least one target affiliated knowledge data; Calculate the innovation coefficient of the target affiliated knowledge data, and sort the target affiliated knowledge data in the data fusion set from largest to smallest according to the innovation coefficient to obtain a target affiliated knowledge data sequence; Generate at least one main fusion node in the main body knowledge data and at least one sub-fusion node in the target affiliated knowledge data; Based on the main fusion node and the sub-fusion node, fuse the target affiliated knowledge data and the main body knowledge data in the data fusion set, and the fusion order is carried out according to the target affiliated knowledge data sequence corresponding to the data fusion set.

[0006] Preferably, the stripping of the complete logical architecture from the data to be merged to obtain at least one complete knowledge data comprises the following steps: Based on the big data, at least one piece of knowledge data including two knowledge points is obtained, each serving as sample knowledge data; The half of the minimum value of the size of the sample knowledge data is used as the preset value; Segment the data to be fused to obtain at least one local knowledge block, where the size of the local knowledge block is a preset value; Based on the big data, at least one application scenario of the local knowledge block is obtained, and features of the at least one application scenario are extracted to obtain at least one application feature; The ratio of the number of overlapping application features of two local knowledge blocks to the total number of application features of the two local knowledge blocks is taken as the correlation degree of the two local knowledge blocks; Calculate the correlation between two sample knowledge data, and use the maximum correlation between the two sample knowledge data without overlapping knowledge points as the correlation threshold; When the correlation between two local knowledge blocks exceeds the correlation threshold, a correlation path is connected between the two local knowledge blocks; Form at least one complete knowledge data, satisfying the following conditions: the complete knowledge data is composed of local knowledge blocks, any two local knowledge blocks in the complete knowledge data are connected by a single-link path, the single-link path is composed of at least one associated path, and there is no associated path between the local knowledge blocks of different complete knowledge data.

[0007] Preferably, the step of classifying the complete knowledge data to form at least one data fusion set comprises the following steps: Count the number of associated paths connecting local knowledge blocks in the complete knowledge data as the characteristic value of the local knowledge block; Sort the local knowledge blocks in the complete knowledge data according to their characteristic values from large to small to obtain a local knowledge block sequence; According to the sequence of local knowledge blocks, the local knowledge blocks are tested for deletion. When the local knowledge blocks are deleted, the complete knowledge data is disconnected. Then the local knowledge blocks are used as the target local knowledge blocks. Connectivity is defined as: any two local knowledge blocks in the complete knowledge data are connected by a single-link path, otherwise, they are disconnected. The target local knowledge blocks are aggregated into main knowledge data, and the local knowledge blocks other than the main knowledge data in the complete knowledge data are respectively used as subsidiary knowledge data.

[0008] Preferably, the step of screening the subsidiary knowledge data in the data fusion set to obtain at least one target subsidiary knowledge data comprises the following steps: After deleting the main knowledge data in the complete knowledge data, at least one affiliated knowledge data link is formed, satisfying that any two affiliated knowledge data in the affiliated knowledge data link are connected by a single-link path, and there is no association path between the affiliated knowledge data in different affiliated knowledge data links; Obtain all possible combinations of two affiliated knowledge data in the affiliated knowledge data link to obtain at least one affiliated knowledge data combination, and the two affiliated knowledge data in the affiliated knowledge data combination are respectively used as the first affiliated knowledge data and the second affiliated knowledge data; Based on big data, obtain at least one first knowledge application event of the first affiliated knowledge data, and obtain the first remaining knowledge used in the first knowledge application event, where the first remaining knowledge does not include the first affiliated knowledge data; Based on big data, obtain at least one second knowledge application event of the second affiliated knowledge data, and obtain the second remaining knowledge used in the second knowledge application event, where the second remaining knowledge does not include the second affiliated knowledge data; Establish a corresponding relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is consistent with the first remaining knowledge, where there is no coverage relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is inconsistent with the first remaining knowledge; Perform deduction and comparison on the second knowledge application event and the first knowledge application event. When the second knowledge application event can be deduced from the first knowledge application event, the second knowledge application event is covered by the first knowledge application event. When the first knowledge application event can be deduced from the second knowledge application event, the first knowledge application event is covered by the second knowledge application event; When the first knowledge application event of the first affiliated knowledge data is respectively covered by the second knowledge application event of the second affiliated knowledge data, delete the first affiliated knowledge data. When the second knowledge application event of the second affiliated knowledge data is covered by the first knowledge application event of the first affiliated knowledge data, delete the second affiliated knowledge data; Respectively use the affiliated knowledge data in the affiliated knowledge data link after the deletion as the target affiliated knowledge data.

[0009] Preferably, the deduction and comparison of the second knowledge application event and the first knowledge application event include the following steps: Obtain at least one sample application requirement, obtain the completed part of the second knowledge application event for the sample application requirement as the second completed part, and obtain the completed part of the first knowledge application event for the sample application requirement as the first completed part; Merge at least one second completed part to obtain a second total part, and merge at least one first completed part to obtain a first total part; Take the intersection of the second general part and the first general part to obtain the characteristic part. Take the part of the second general part that is different from the characteristic part as the second target part, and take the part of the first general part that is different from the characteristic part as the first target part; Based on big data, obtain at least one common sense requirement; When the second target part is an empty set or a common sense requirement, the second knowledge application event can be deduced from the first knowledge application event. When the first target part is an empty set or a common sense requirement, the first knowledge application event can be deduced from the second knowledge application event.

[0010] Preferably, the obtaining at least one common sense requirement based on big data includes the following steps: Based on big data, obtain at least one knowledge application requirement; Form a set of sample testers, which is composed of personnel in various occupations in equal proportions; When the knowledge application requirements can be completed by the personnel in the set of sample testers without referring to materials, then the knowledge application requirements are regarded as common sense requirements.

[0011] Preferably, the calculating the innovation coefficient of the target affiliated knowledge data includes the following steps: In big data, obtain the time when the target affiliated knowledge data first appears as the target time; Obtain the first processing result of at least one actual application requirement using the target affiliated knowledge data, and obtain the second processing result of the actual application requirement before the target time; Statistically analyze the first processing time consumed by the first processing result, and statistically analyze the second processing time consumed by the second processing result. Divide the second processing time by the first processing time to obtain the first coefficient; Statistically analyze the first completion degree of the first processing result for the actual application requirement, and statistically analyze the second completion degree of the second processing result for the actual application requirement. Divide the first completion degree by the second completion degree to obtain the second coefficient; Multiply the first coefficient by the second coefficient to obtain the innovation coefficient.

[0012] Preferably, the generating at least one main fusion node in the main body knowledge data and generating at least one secondary fusion node in the target affiliated knowledge data includes the following steps: Segment the local knowledge blocks in the main body knowledge data to obtain at least one main fusion node, and the knowledge in the main fusion node corresponds to the same entity; Segment the target affiliated knowledge data to obtain at least one secondary fusion node, and the knowledge in the secondary fusion node corresponds to the same entity; Establish a corresponding relationship between the main fusion node and the secondary fusion node corresponding to the same entity.

[0013] Preferably, the fusion of the target affiliated knowledge data and the main body knowledge data in the data fusion set includes the following steps: Obtain at least one main application event of the knowledge in the main fusion node, and obtain at least one secondary application event of the knowledge in the secondary fusion node; Deduce and compare the main application event and the secondary application event. When the main application event can be deduced from the secondary application event, the fusion result of the main fusion node and the secondary fusion node is the secondary fusion node. When the secondary application event can be deduced from the main application event, the fusion result of the main fusion node and the fusion node is the main fusion node; When the main application event and the secondary application event cannot be deduced from each other, take the intersection of the knowledge in the main fusion node and the knowledge in the secondary fusion node to obtain the first fusion part; Take the part of the knowledge in the main fusion node that is different from the first fusion part as the main category part; Take the part of the knowledge in the secondary fusion node that is different from the first fusion part as the secondary category part; Take the conflicting parts in the main category part and the secondary category part as the main conflict part and the secondary conflict part respectively. The main conflict part belongs to the main category part, and the secondary conflict part belongs to the secondary category part; Fuse the main category part and the secondary category part. When fusing, delete the secondary category part to obtain the second fusion part; Summarize the first fusion part and the second fusion part to complete the fusion.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: By stripping the complete logical architecture in the data to be fused, classifying the complete knowledge data, screening the affiliated knowledge data in the data fusion set, and fusing the target affiliated knowledge data and the main body knowledge data in the data fusion set, a complete knowledge architecture can be obtained by means of data stripping, and the attributes of the knowledge can be judged according to the application situation of the knowledge, so as to delete the data to be fused, reduce the workload of subsequent processing. Using the main body knowledge data and the affiliated knowledge data, the main body knowledge data can be used as the main body architecture for subsequent fusion, making the fusion more orderly. Through the fusion processing of the main fusion node and the secondary fusion node, the conflicting situations can be handled during fusion, making the final fusion effect more reasonable, and thus meeting the actual needs. Description of the Drawings

[0015] Figure 1 It is a schematic flowchart of the knowledge data fusion method based on artificial intelligence technology mixing of the present invention; Figure 2Schematic diagram of the process of stripping the complete logical structure from the data to be fused in the present invention to obtain at least one complete knowledge data; Figure 3 Schematic diagram of the process of classifying the complete knowledge data in the present invention to form at least one data fusion set; Figure 4 Schematic diagram of the process of screening the affiliated knowledge data in the data fusion set in the present invention to obtain at least one target affiliated knowledge data; Figure 5 Schematic diagram of the process of deducing and comparing the second knowledge application event and the first knowledge application event in the present invention; Figure 6 Schematic diagram of the process of obtaining at least one common sense requirement based on big data in the present invention; Figure 7 Schematic diagram of the process of calculating the innovation coefficient of the target affiliated knowledge data in the present invention; Figure 8 Schematic diagram of the process of generating at least one main fusion node in the main body knowledge data and at least one secondary fusion node in the target affiliated knowledge data in the present invention; Figure 9 Schematic diagram of the process of fusing the target affiliated knowledge data and the main body knowledge data in the data fusion set in the present invention. Detailed implementation manners

[0016] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0017] Refer to Figure 1 As shown, a knowledge data fusion method based on a hybrid of artificial intelligence technologies includes: Summarize the knowledge data to be fused into data to be fused, strip the complete logical structure from the data to be fused to obtain at least one complete knowledge data; Classify the complete knowledge data to form at least one data fusion set, and the data fusion set is composed of main body knowledge data and at least one affiliated knowledge data; Screen the affiliated knowledge data in the data fusion set to obtain at least one target affiliated knowledge data; Calculate the innovation coefficient of the target affiliated knowledge data, and sort the target affiliated knowledge data in the data fusion set from largest to smallest according to the innovation coefficient to obtain a target affiliated knowledge data sequence; Generate at least one main fusion node in the main body knowledge data and at least one secondary fusion node in the target affiliated knowledge data; Based on the main fusion node and the secondary fusion node, the target affiliated knowledge data and the main body knowledge data in the data fusion set are fused, and the fusion order is carried out according to the target affiliated knowledge data sequence corresponding to the data fusion set.

[0018] There are various data interleaved settings in the data to be fused. Therefore, during fusion, different types of data cannot be fused, and fusion needs to be carried out among the same type of data. Therefore, it is necessary to strip the data in the data to be fused to obtain complete knowledge data. Each complete knowledge data contains the same type of data that can be fused. However, the complete knowledge data involves a lot of knowledge, and there may be approximate knowledge, but their innovation degrees are different, and there may also be contradictions. If these situations are not handled, it is easy to lead to a too high redundancy degree of the fused data, which imposes a great burden on subsequent use. Therefore, corresponding steps are set in this solution to handle this situation.

[0019] Refer to Figure 2 As shown, the steps to strip the complete logical structure in the data to be fused to obtain at least one complete knowledge data include the following: Based on big data, obtain at least one knowledge data containing two knowledge points, and use them as sample knowledge data respectively; Take half of the minimum value of the size of the sample knowledge data as the preset value; Divide the data to be fused to obtain at least one local knowledge block, and the size of the local knowledge block is the preset value; Based on big data, obtain at least one application scenario of the local knowledge block, and extract features from at least one application scenario to obtain at least one application feature; Take the ratio of the number of overlapping application features of two local knowledge blocks to the total number of application features of the two local knowledge blocks as the correlation degree of the two local knowledge blocks; Calculate the correlation degree of two sample knowledge data, and take the maximum value of the correlation degrees of two sample knowledge data without overlapping knowledge points as the correlation threshold; When the correlation degree of two local knowledge blocks exceeds the correlation threshold, connect a correlation path between the two local knowledge blocks; Form at least one complete knowledge data, satisfying that: the complete knowledge data is composed of local knowledge blocks, any two local knowledge blocks in the complete knowledge data are connected by a single-link path, the single-link path is composed of at least one correlation path, and there is no correlation path between the local knowledge blocks of different complete knowledge data.

[0020] When performing stripping, associated paths are used to connect local knowledge blocks. When there is an associated path connecting two local knowledge blocks, it indicates that there is a certain correlation between them. Therefore, any two local knowledge blocks in the complete knowledge data formed thereby can be conducted through multiple associated paths via other local knowledge blocks. Since the local knowledge blocks connected by associated paths have a certain correlation, the local knowledge blocks in the complete knowledge data are correlated. And because there is no associated path between the local knowledge blocks in different complete knowledge data, the local knowledge blocks in different complete knowledge data have no correlation. Therefore, each complete knowledge data is an independent whole, and thus targeted knowledge fusion can be performed in the complete knowledge data.

[0021] Refer to Figure 3 As shown, the steps for classifying the complete knowledge data into at least one data fusion set include the following: Count the number of associated paths connected by local knowledge blocks in the complete knowledge data as the eigenvalue of the local knowledge block; Sort the local knowledge blocks in the complete knowledge data in descending order of the eigenvalue to obtain a local knowledge block sequence; According to the order of the local knowledge block sequence, perform a deletion test on the local knowledge blocks. When a local knowledge block is deleted and the complete knowledge data is not conducted, then this local knowledge block is regarded as the target local knowledge block. The definition of conduction is that any two local knowledge blocks in the complete knowledge data are connected by a single connection path; otherwise, it is not conducted; Summarize the target local knowledge blocks as the main body knowledge data, and regard the local knowledge blocks in the complete knowledge data other than the main body knowledge data as the affiliated knowledge data respectively.

[0022] Since the data contained in the complete knowledge data may be numerous, a main body framework needs to be formed for fusion during fusion. Here, the main body framework is formed by the main body knowledge data, and the affiliated knowledge data is fused as an attachment into the main body framework. The identification of the main body knowledge data mainly depends on its importance in the complete knowledge data. The main body knowledge data is composed of important local knowledge blocks. The identification of important local knowledge blocks mainly depends on the associated paths connected to the local knowledge blocks. When a local knowledge block is removed, it will cause two local knowledge blocks in the complete knowledge data to be unable to be connected by a single connection path because the single connection path originally used to connect two local knowledge blocks passes through the removed local knowledge block. When this local knowledge block is removed, this single connection path is broken. Thus, it shows that the removed local knowledge block is relatively important because it is not the relatively marginal data in the correlation. Otherwise, its removal will not affect the conduction of the complete knowledge data.

[0023] Refer to Figure 4As shown in the figure, screening the subsidiary knowledge data in the data fusion set to obtain at least one target subsidiary knowledge data includes the following steps: After deleting the main body knowledge data in the complete knowledge data, at least one subsidiary knowledge data link is formed, satisfying that any two subsidiary knowledge data in the subsidiary knowledge data link are connected by a single-link path, and there is no association path between the subsidiary knowledge data in different subsidiary knowledge data links; Obtain all possible combinations of two subsidiary knowledge data in the subsidiary knowledge data link to obtain at least one subsidiary knowledge data combination, and the two subsidiary knowledge data in the subsidiary knowledge data combination are respectively used as the first subsidiary knowledge data and the second subsidiary knowledge data; Based on big data, obtain at least one first knowledge application event of the first subsidiary knowledge data, and obtain the first remaining knowledge used in the first knowledge application event, where the first remaining knowledge does not include the first subsidiary knowledge data; Based on big data, obtain at least one second knowledge application event of the second subsidiary knowledge data, and obtain the second remaining knowledge used in the second knowledge application event, where the second remaining knowledge does not include the second subsidiary knowledge data; Establish a corresponding relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is consistent with the first remaining knowledge. Among them, there is no coverage relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is inconsistent with the first remaining knowledge; Carry out deduction and comparison on the second knowledge application event and the first knowledge application event. When the second knowledge application event can be deduced from the first knowledge application event, the second knowledge application event is covered by the first knowledge application event. When the first knowledge application event can be deduced from the second knowledge application event, the first knowledge application event is covered by the second knowledge application event; When the first knowledge application event of the first subsidiary knowledge data is respectively covered by the second knowledge application event of the second subsidiary knowledge data, the first subsidiary knowledge data is deleted. When the second knowledge application event of the second subsidiary knowledge data is covered by the first knowledge application event of the first subsidiary knowledge data, the second subsidiary knowledge data is deleted; Respectively use the subsidiary knowledge data in the subsidiary knowledge data link after the deletion as the target subsidiary knowledge data.

[0024] There are many subsidiary knowledge data in the complete knowledge data, but there may be replaceable situations between the subsidiary knowledge data. For example, two subsidiary knowledge data are both for the same entity, but one is more advanced and the other is more backward, and the backward one is included by the advanced one. For similar situations, only one of them needs to be retained. Thus, the subsequent fusion workload can be reduced.

[0025] Refer to Figure 5As shown in the figure, the deduction and comparison of the second knowledge application event and the first knowledge application event include the following steps: Obtain at least one sample application requirement, obtain the completed part of the second knowledge application event for the sample application requirement as the second completed part, and obtain the completed part of the first knowledge application event for the sample application requirement as the first completed part; Merge at least one second completed part to obtain a second total part, and merge at least one first completed part to obtain a first total part; Take the intersection of the second total part and the first total part to obtain a characteristic part. Take the part of the second total part that is different from the characteristic part as the second target part, and take the part of the first total part that is different from the characteristic part as the first target part; Based on big data, obtain at least one common sense requirement; When the second target part is an empty set or a common sense requirement, the second knowledge application event can be deduced from the first knowledge application event. When the first target part is an empty set or a common sense requirement, the first knowledge application event can be deduced from the second knowledge application event.

[0026] When comparing approximate knowledge, it is difficult to directly obtain the relationship between them. When two pieces of knowledge can be deduced, they can be deleted. Therefore, the deduced relationship can determine whether one of the two pieces of knowledge can be used for replacement. And because the deleted knowledge is deduced from the knowledge that can be retained, the fusion effect of using the retained knowledge is the same as that before deletion and fusion. The deduction principle depends on the comparison of the first knowledge application event of the first affiliated knowledge data and the second knowledge application event of the second affiliated knowledge data. When comparing, use the sample application requirement to analyze the completion situation of the first knowledge application event and the second knowledge application event. When analyzing, when the second target part is an empty set or a common sense requirement, it means that the second target part can be completed by any person without assistance according to common sense. Therefore, since the second target part is the different part of the second total part from the first total part, the second total part can be obtained from common sense and the first total part. Therefore, the deduction relationship between the second knowledge application event and the first knowledge application event can be determined. The same analysis applies to the case where the first target part is an empty set or a common sense requirement.

[0027] Refer to Figure 6 As shown in the figure, based on big data, obtaining at least one common sense requirement includes the following steps: Based on big data, obtain at least one knowledge application requirement; Form a sample tester set, and the sample tester set is composed of an equal proportion of personnel in various occupations; When the knowledge application requirements are all completed by the personnel in the sample tester set without referring to materials, the knowledge application requirements are regarded as common sense requirements.

[0028] Referring to Figure 7 as shown, calculating the innovation coefficient of the target affiliated knowledge data includes the following steps: In big data, obtain the time when the target affiliated knowledge data first appears as the target time; Obtain the first processing result of at least one actual application requirement using the target affiliated knowledge data, and obtain the second processing result of the actual application requirement before the target time; Statistically analyze the first processing time-consuming of the first processing result, statistically analyze the second processing time-consuming of the second processing result, divide the second processing time-consuming by the first processing time-consuming to obtain the first coefficient; Statistically analyze the first completion degree of the first processing result for the actual application requirement, statistically analyze the second completion degree of the second processing result for the actual application requirement, divide the first completion degree by the second completion degree to obtain the second coefficient; Multiply the first coefficient by the second coefficient to obtain the innovation coefficient.

[0029] The innovation coefficient actually depicts the optimization degree of the target affiliated knowledge data compared with the previous data, mainly depicted from the time and completion degree of whether to use the target affiliated knowledge data for processing. When the target affiliated knowledge data is more optimized than the previous data, its completion degree for the actual application requirement is higher. At the same time, its processing time will be shorter. Therefore, divide the second processing time-consuming by the first processing time-consuming to obtain the first coefficient, divide the first completion degree by the second completion degree to obtain the second coefficient. When the target affiliated knowledge data is more optimized than the previous data, both the first coefficient and the second coefficient will be larger. Thus, the innovation coefficient will also be larger. Therefore, the innovation coefficient is positively correlated with the optimization situation of the target affiliated knowledge data. Therefore, the innovation coefficient can depict the optimization situation of the target affiliated knowledge data.

[0030] Referring to Figure 8 as shown, generating at least one main fusion node in the main body knowledge data and generating at least one secondary fusion node in the target affiliated knowledge data includes the following steps: Segment the local knowledge blocks in the main body knowledge data to obtain at least one main fusion node, and the knowledge in the main fusion node corresponds to the same entity; Segment the target affiliated knowledge data to obtain at least one secondary fusion node, and the knowledge in the secondary fusion node corresponds to the same entity; Establish a corresponding relationship between the main fusion node and the secondary fusion node corresponding to the same entity.

[0031] Each piece of knowledge corresponds to an entity, that is, the thing it describes. For example, a physical equation must correspond to a certain entity, which may be a gas, a fluid, or a solid, or it may be an application object. The application object is the sum of all application objects to which the knowledge is applied.

[0032] Refer to Figure 9 As shown, the fusion of the target affiliated knowledge data and the main knowledge data in the data fusion set includes the following steps: Obtain at least one main application event of the knowledge in the main fusion node, and obtain at least one secondary application event of the knowledge in the secondary fusion node; Deduce and compare the main application event and the secondary application event. When the main application event can be deduced from the secondary application event, the fusion result of the main fusion node and the secondary fusion node is the secondary fusion node. When the secondary application event can be deduced from the main application event, the fusion result of the main fusion node and the fusion node is the main fusion node; When the main application event and the secondary application event cannot be deduced from each other, take the intersection of the knowledge in the main fusion node and the knowledge in the secondary fusion node to obtain the first fusion part; Take the part of the knowledge in the main fusion node that is different from the first fusion part as the main category part; Take the part of the knowledge in the secondary fusion node that is different from the first fusion part as the secondary category part; Take the conflicting parts in the main category part and the secondary category part as the main conflict part and the secondary conflict part respectively. The main conflict part belongs to the main category part, and the secondary conflict part belongs to the secondary category part; Fuse the main category part and the secondary category part. When fusing, delete the secondary category part to obtain the second fusion part; Summarize the first fusion part and the second fusion part to complete the fusion.

[0033] When fusing, it is necessary to consider approximate knowledge and conflicting situations. Approximate knowledge can be determined by deduction. For knowledge that cannot be deduced from each other, the non-conflicting parts can be merged, and for the conflicting parts, the corresponding parts of the knowledge in the main fusion node are used because the knowledge in the main fusion node is more important and the credibility of its data is higher.

[0034] Furthermore, this solution also proposes a storage medium on which a computer-readable program is stored. When the computer-readable program is called, it executes the above-mentioned knowledge data fusion method based on artificial intelligence technology hybridization.

[0035] It can be understood that the storage medium can be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid-state drive (SSD).

[0036] In summary, the advantages of the present invention are as follows: by stripping the complete logical architecture from the data to be fused, classifying the complete knowledge data, screening the attached knowledge data in the data fusion set, and fusing the target attached knowledge data and the main body knowledge data in the data fusion set, a complete knowledge architecture can be obtained through data stripping, and the attributes of the knowledge can be judged according to the application situation of the knowledge, so as to delete the data waiting to be fused, reduce the workload of subsequent processing. Using the main body knowledge data and the attached knowledge data, the main body knowledge data can be used as the main body architecture for subsequent fusion, making the fusion more orderly. Through the fusion processing of the main fusion node and the secondary fusion node, the contradictions can be handled during the fusion, making the final fusion effect more reasonable, and thus meeting the actual requirements.

[0037] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A knowledge data fusion method based on the hybrid of artificial intelligence technologies, characterized in that, Including: Summarize the knowledge data waiting for fusion into data to be fused, strip the complete logical architecture from the data to be fused, and obtain at least one complete knowledge data; Classify the complete knowledge data to form at least one data fusion set, and the data fusion set consists of main body knowledge data and at least one affiliated knowledge data; Screen the affiliated knowledge data in the data fusion set to obtain at least one target affiliated knowledge data; Calculate the innovation coefficient of the target affiliated knowledge data, sort the target affiliated knowledge data in the data fusion set from largest to smallest according to the innovation coefficient, and obtain the target affiliated knowledge data sequence; Generate at least one main fusion node in the main body knowledge data and at least one sub-fusion node in the target affiliated knowledge data; Based on the main fusion node and the sub-fusion node, fuse the target affiliated knowledge data and the main body knowledge data in the data fusion set, and the fusion order is carried out according to the target affiliated knowledge data sequence corresponding to the data fusion set.

2. The knowledge data fusion method based on the hybrid of artificial intelligence technology according to claim 1, wherein The step of stripping the complete logical architecture from the data to be fused to obtain at least one complete knowledge data includes the following steps: Based on big data, obtain at least one knowledge data containing two knowledge points, and use them as sample knowledge data respectively; Take half of the minimum value of the size of the sample knowledge data as the preset value; Divide the data to be fused to obtain at least one local knowledge block, and the size of the local knowledge block is the preset value; Based on big data, obtain at least one application scenario of the local knowledge block, extract features from at least one application scenario, and obtain at least one application feature; Take the ratio of the number of overlapping application features of two local knowledge blocks to the total number of application features of the two local knowledge blocks as the correlation degree of the two local knowledge blocks; Calculate the correlation degree of two sample knowledge data, and take the maximum value of the correlation degrees of two sample knowledge data without overlapping knowledge points as the correlation threshold; When the correlation degree of two local knowledge blocks exceeds the correlation threshold, connect an association path between the two local knowledge blocks; Form at least one complete knowledge data, satisfying that the complete knowledge data is composed of local knowledge blocks, any two local knowledge blocks in the complete knowledge data are connected by a single-link path, the single-link path is composed of at least one association path, and there is no association path between the local knowledge blocks of different complete knowledge data.

3. A method for fusing knowledge data based on artificial intelligence technology mixture according to claim 2, characterized in that The step of classifying the complete knowledge data to form at least one data fusion set includes the following steps: Count the number of association paths connected by local knowledge blocks in the complete knowledge data as the characteristic value of the local knowledge block; Sort the local knowledge blocks in the complete knowledge data from largest to smallest according to the characteristic value to obtain the local knowledge block sequence; According to the order of the local knowledge block sequence, conduct a deletion test on the local knowledge block. When the local knowledge block is deleted and the complete knowledge data is not conductive, then regard the local knowledge block as the target local knowledge block. The definition of conduction is that any two local knowledge blocks in the complete knowledge data are connected by a single-link path, otherwise, it is not conductive; Summarize the target local knowledge chunks into the main body knowledge data, and use the local knowledge chunks other than the main body knowledge data in the complete knowledge data as the affiliated knowledge data respectively.

4. A method for fusing knowledge data based on artificial intelligence technology hybridization according to claim 3, characterized in that, The steps of screening the affiliated knowledge data in the data fusion set to obtain at least one target affiliated knowledge data include the following: After deleting the main body knowledge data in the complete knowledge data, at least one affiliated knowledge data link is formed, satisfying that any two affiliated knowledge data in the affiliated knowledge data link are connected by a single-link path, and there is no association path between the affiliated knowledge data in different affiliated knowledge data links; Obtain all possible combinations of two affiliated knowledge data in the affiliated knowledge data link to obtain at least one affiliated knowledge data combination, and use the two affiliated knowledge data in the affiliated knowledge data combination as the first affiliated knowledge data and the second affiliated knowledge data respectively; Based on big data, obtain at least one first knowledge application event of the first affiliated knowledge data, and obtain the first remaining knowledge used in the first knowledge application event, where the first remaining knowledge does not include the first affiliated knowledge data; Based on big data, obtain at least one second knowledge application event of the second affiliated knowledge data, and obtain the second remaining knowledge used in the second knowledge application event, where the second remaining knowledge does not include the second affiliated knowledge data; Establish a corresponding relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is consistent with the first remaining knowledge. Among them, there is no coverage relationship between the second knowledge application event and the first knowledge application event where the second remaining knowledge is inconsistent with the first remaining knowledge; Perform deduction and comparison on the second knowledge application event and the first knowledge application event. When the second knowledge application event can be deduced from the first knowledge application event, the second knowledge application event is covered by the first knowledge application event. When the first knowledge application event can be deduced from the second knowledge application event, the first knowledge application event is covered by the second knowledge application event; When the first knowledge application event of the first affiliated knowledge data is covered by the second knowledge application event of the second affiliated knowledge data respectively, delete the first affiliated knowledge data. When the second knowledge application event of the second affiliated knowledge data is covered by the first knowledge application event of the first affiliated knowledge data, delete the second affiliated knowledge data; Use the affiliated knowledge data in the affiliated knowledge data link after the deletion is completed as the target affiliated knowledge data respectively.

5. A method for fusing knowledge data based on artificial intelligence technology mixture according to claim 4, characterized in that, The steps of performing deduction and comparison on the second knowledge application event and the first knowledge application event include the following: Obtain at least one sample application requirement, obtain the completed part of the second knowledge application event for the sample application requirement as the second completed part, and obtain the completed part of the first knowledge application event for the sample application requirement as the first completed part; Merge at least one second completed part to obtain the second total part, and merge at least one first completed part to obtain the first total part; Take the intersection of the second total part and the first total part to obtain the characteristic part, use the part of the second total part different from the characteristic part as the second target part, and use the part of the first total part different from the characteristic part as the first target part; Based on big data, obtain at least one common sense requirement; When the second target part is an empty set or a common sense requirement, the second knowledge application event can be deduced from the first knowledge application event. When the first target part is an empty set or a common sense requirement, the first knowledge application event can be deduced from the second knowledge application event.

6. A knowledge data fusion method based on artificial intelligence technology hybridization according to claim 5, characterized in that The obtaining of at least one common sense requirement based on big data includes the following steps: Based on big data, obtain at least one knowledge application requirement; Form a set of sample testers, and the set of sample testers is composed of personnel engaged in various occupations in equal proportions; When the knowledge application requirements can be completed by the personnel in the set of sample testers without referring to materials, then the knowledge application requirements are regarded as common sense requirements.

7. A knowledge data fusion method based on the hybrid of artificial intelligence technology according to claim 6, characterized in that, The calculation of the innovation coefficient of the target affiliated knowledge data includes the following steps: In big data, obtain the time when the target affiliated knowledge data first appears as the target time; Obtain the first processing result of at least one actual application requirement using the target affiliated knowledge data, and obtain the second processing result of the actual application requirement before the target time; Statistically analyze the first processing time consumed by the first processing result, and statistically analyze the second processing time consumed by the second processing result. Divide the second processing time by the first processing time to obtain the first coefficient; Statistically analyze the first completion degree of the first processing result for the actual application requirement, and statistically analyze the second completion degree of the second processing result for the actual application requirement. Divide the first completion degree by the second completion degree to obtain the second coefficient; Multiply the first coefficient by the second coefficient to obtain the innovation coefficient.

8. A method for fusing knowledge data based on artificial intelligence technology hybridization according to claim 7, characterized in that The generation of at least one main fusion node in the main body knowledge data and the generation of at least one sub-fusion node in the target affiliated knowledge data include the following steps: Segment the local knowledge blocks in the main body knowledge data to obtain at least one main fusion node, and the knowledge in the main fusion node corresponds to the same entity; Segment the target affiliated knowledge data to obtain at least one sub-fusion node, and the knowledge in the sub-fusion node corresponds to the same entity; Establish a corresponding relationship between the main fusion node and the sub-fusion node corresponding to the same entity.

9. A knowledge data fusion method based on the hybrid of artificial intelligence technology according to claim 8, characterized in that, The fusion of the target affiliated knowledge data and the main body knowledge data in the data fusion set includes the following steps: Obtain at least one main application event of the knowledge in the main fusion node, and obtain at least one sub-application event of the knowledge in the sub-fusion node; Deduce and compare the main application event and the sub-application event. When the main application event can be deduced from the sub-application event, the fusion result of the main fusion node and the sub-fusion node is the sub-fusion node. When the sub-application event can be deduced from the main application event, the fusion result of the main fusion node and the fusion node is the main fusion node; When the main application event and the sub-application event cannot be deduced from each other, take the intersection of the knowledge in the main fusion node and the knowledge in the sub-fusion node to obtain the first fusion part; Take the part of the knowledge in the main fusion node that is different from the first fusion part as the main category part; Take the part of the knowledge in the sub-fusion node that is different from the first fusion part as the sub-category part; The parts in contradiction in the main category part and the sub-category part are respectively taken as the main contradiction part and the sub-contradiction part. The main contradiction part belongs to the main category part, and the sub-contradiction part belongs to the sub-category part; Fuse the main category part and the sub-category part. When fusing, delete the sub-category part to obtain the second fused part; Summarize the first fused part and the second fused part to complete the fusion.

Images