Knowledge clustering method oriented to subversive concept innovation

Abstract

The invention relates to a concept innovation-oriented analogy knowledge clustering method. The method comprises the following steps of: 1, determining a function similarity key characteristic index system of knowledge clustering; 2, performing knowledge retrieval according to the key function indexes; 3, calculating the measurement of the functional similarity between the clustering knowledge according to a cosine law method, and setting a threshold value according to a calculation result; 4, calculating a technical difference degree parameter of each piece of knowledge to be clustered; and step 5, completing clustering of knowledge of various ratios through calculation results of the two clustering indexes, namely the function similarity and the technical difference. According to the cross-domain knowledge clustering method, efficient cross-domain knowledge clustering can be effectively achieved, analogy knowledge in different fields can be effectively processed through the clustering method combining the two dimensions of function similarity and technical difference, efficient clustering of cross-domain knowledge is achieved, and generation of higher-quality subversive innovation concepts is facilitated.

Description

Technical Field

[0001] The present invention belongs to the field of computer data processing, and particularly relates to a knowledge clustering method for disruptive concept innovation. Background Art

[0002] With the rapid development of generative artificial intelligence technology, there is a lack of an efficient and reliable support method for further effectively enhancing the data processing advantage of computers in innovative design, especially in the generation of disruptive innovation concepts, and effectively helping designers to form innovative solutions. The generation of disruptive innovation concepts relies on the analogy and application of cross-domain principles and knowledge. Therefore, how to utilize the efficient information processing ability of computers, face the specific design tasks of disruptive concept innovation, and by means of scientific methods, identify the most critical analogical knowledge from a vast amount of available innovation knowledge, and achieve effective clustering of available analogical knowledge according to its different characteristic dimensions, can provide designers with practical and highly valuable innovation knowledge, can significantly improve the efficiency of concept innovation, especially has important practical significance in inspiring the generation of disruptive innovation concept solutions.

[0003] Currently, in the extensive practice of concept innovation design, especially in the process of generating disruptive innovation concepts, cross-domain knowledge needs to be applied. However, currently, in the process of selecting analogical innovation knowledge, it mainly relies on domain experts to give subjective evaluations based on their professional knowledge and experience. The generation of disruptive innovation concepts mainly depends on the personal knowledge reserve, personal specialty, and interest preferences of designers, which will increase the uncertainty of disruptive innovation to a certain extent. Especially when there are many available similar and homogeneous analogical knowledge, the personal professional limitations and preferences of experts will have an adverse impact on the objectivity of the selection and application process of analogical knowledge, and this deviation will be continuously amplified in the subsequent more specific detailed design process, resulting in an adverse effect on the efficiency of disruptive concept innovation. To solve the problem in the current process of disruptive concept innovation, especially to generate high-quality disruptive innovation concepts by using cross-domain highly relevant analogical knowledge, to achieve efficient knowledge clustering of analogical knowledge from different sources and backgrounds, and to avoid over-relying on the personal preferences of scoring experts in the clustering process of heterogeneous analogical knowledge, resulting in efficient knowledge management in the process of generating disruptive innovation concepts, a knowledge clustering method for generating disruptive innovation concepts based on two dimensions of functional feature similarity and technical system difference degree is proposed, which makes up for the deficiencies of the current method and fills the method gap in this application field. Summary of the Invention

[0004] The present invention aims at the deficiencies of the prior art and proposes a knowledge clustering method for disruptive concept innovation.

[0005] The above object of the present invention is achieved by the following technical solutions:

[0006] A knowledge clustering method for disruptive concept innovation, comprising the following steps:

[0007] Step 1: Determine the key feature index system of functional similarity for knowledge clustering;

[0008] Step 2: Conduct knowledge retrieval according to the key function indicators;

[0009] Step 3: Calculate the measure of functional similarity between each clustering knowledge according to the cosine theorem method, and set a threshold according to the calculation result;

[0010] Step 4: Calculate the technical difference degree parameters of each knowledge to be clustered;

[0011] Step 5: Complete the clustering of various analogous knowledge through the calculation results of two clustering indicators, namely functional similarity and technical difference degree, where functional similarity and technical difference degree constitute two knowledge clustering dimensions.

[0012] Moreover, in Step 1, the key feature index system of functional similarity is the specific content of each index of the key functional features of analogous innovation knowledge, and the total number of functional feature indicators is determined, denoted as n;

[0013] Moreover, Step 3 includes:

[0014] 3.1. Select the quantization criterion for the functional feature similarity of the synonym cluster, and use the quantization criterion to calculate the weight values of each functional feature in the functional feature sequence. The weight values of the n functional features constitute a weight vector:

[0015] w = (w1, w2, …, w n ) T ;

[0016] 3.2. Use the cosine theorem based on trigonometric function features to calculate the similarity of functional features determined by the synonym cluster in the analogous knowledge unit, including the following two scenarios:

[0017] Scenario 1: Using the relationship in the synonym cluster, if two words are not on the same tree, the similarity is recorded as: Sim(A, B) = f

[0018] Scenario 2: If two words are on the same tree

[0019] If at the second-level branch, the coefficient is a, and the similarity between the two is expressed by the following formula:

[0020]

[0021] If it branches at the 3rd layer with a coefficient of b, the similarity between the two is expressed by the following formula:

[0022]

[0023] If it branches at the 4th layer with a coefficient of c, the similarity between the two is expressed by the following formula:

[0024]

[0025] If it branches at the 5th layer with a coefficient of d, the similarity between the two is expressed by the following formula:

[0026]

[0027] Among them, (n - k + 1) / n is a control parameter, where n is the total number of nodes in the branch layer and k is the distance between the two branches; if the numbers of the two words are the same and the word-ending flag is "=", the similarity is 1, and if the word-ending symbol is "#", the similarity is e; the initial values of the layer numbers are set as: a = 0.65, b = 0.8, c = 0.9, d = 0.96, e = 0.5, f = 0.1.

[0028] 3.3. Obtain the similarity between feature vectors through the cosine theorem to determine the similarity degree between concepts, specifically:

[0029] The cosine expression of the angle between two n-dimensional vectors X(x1, x2…, xn) and Y(y1, y2…, yn) is as follows:

[0030]

[0031] The similarity between two concepts represents the cosine of n-dimensional Euclidean space, also known as Euclidean space cosine, as shown in the following formula.

[0032]

[0033] Moreover, in step 4, it includes:

[0034] 4.1. First, confirm the quantification of technical difference degree from the perspective of system composition; among them, the components of the technical system include four parts: the execution system, the transmission system, the power system, and the control system;

[0035] 4.2. Then, through the method of assigning weights and normalizing, form the feature vector C describing the function implementation background of the system, C = [Co, Ct, Ce, Cc], and the four parameters in this vector respectively correspond to the technical difference degrees of the execution system, the transmission system, the power system, and the control system;

[0036] 4.3. Respectively take the ratios of 10:6:3:1 as the degrees of difference at the physical principle layer, working principle layer, embodiment level, and detail level, and quantify them into [1, 0.6, 0.3, 0.1] according to the normalization requirements;

[0037] 4.4. Compare the knowledge to be clustered retrieved in step 2 with the existing technical system solutions, confirm the specific functional implementation background feature vector values, and perform normalization processing to calculate the specific values of the technical system difference degree.

[0038] Moreover, in step 5, based on the calculation results of the comprehensive analogical knowledge in two dimensions of functional feature similarity and technical system difference degree, establish a coordinate system for the analogical innovation knowledge that meets the calculation threshold and has potential available value according to the two dimensions; among them, the measurement value of the functional feature is used as the abscissa of the clustering result, and the difference degree of the technical system is used as the ordinate of the clustering result; then each knowledge unit to be clustered realizes the specific coordinates and specific positions of different analogical knowledge in the clustering system according to the two dimensions of its functional feature parameters and technical system difference degree, and completes the clustering of the analogical innovation knowledge.

[0039] The advantages and positive effects of the present invention are as follows:

[0040] The clustering method proposed in this patent fills the gap in the current technical field, makes up for the deficiencies of the existing methods in the face of heterogeneous analogical knowledge, and provides a systematic and scientific solution. First, the knowledge clustering method based on functional feature similarity and technical system difference degree can efficiently identify key analogical knowledge from a large amount of innovative knowledge, provide more practical innovative solutions for designers, and significantly improve the generation efficiency of disruptive innovation concepts. On the other hand, this method avoids the limitations of traditional reliance on expert subjective evaluation. Especially when selecting and applying analogical knowledge, it can reduce the deviation of expert personal preferences and knowledge reserves, ensuring that the clustering results are more objective and scientific; it can effectively achieve efficient cross-domain knowledge clustering. By combining the clustering methods of two dimensions of functional similarity and technical difference degree, the present invention can effectively process analogical knowledge in different fields and achieve efficient cross-domain knowledge clustering, which helps to generate higher-quality disruptive innovation concepts. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart of an analogical knowledge clustering method for disruptive concept innovation;

[0042] Figure 2 It is each relevant functional module of an analogical knowledge clustering method for disruptive concept innovation;

[0043] Figure 3 It is an analysis schematic diagram of an embodiment of an analogical knowledge clustering method for disruptive concept innovation;

[0044] Figure 4 It is a clustering schematic diagram of an embodiment of an analogical knowledge clustering method for disruptive concept innovation;

[0045] Figure 5 It is a schematic diagram of a system of an analogical knowledge clustering method for disruptive concept innovation; Detailed implementation manners

[0046] The structure of the present invention will be further described below with reference to the accompanying drawings and through embodiments. It should be noted that this embodiment is narrative rather than restrictive.

[0047] An analogical knowledge clustering method for disruptive concept innovation, please refer to Figures 1-5 , and its inventive point is that it includes the following steps:

[0048] Step 1: Determine the key feature index system of functional similarity for knowledge clustering;

[0049] Step 2: Perform knowledge retrieval according to the key functional indicators;

[0050] Step 3: Calculate the measure of functional similarity between each clustering knowledge according to the cosine theorem method, and set a threshold according to the calculation result; generally, the threshold of functional similarity is set to 0.3.

[0051] Step 4: Calculate the technical difference degree parameters of each knowledge to be clustered;

[0052] Step 5: Complete the clustering of various analogical knowledge through the calculation results of two clustering indicators of functional similarity and technical difference degree, where the functional similarity and technical difference degree constitute two knowledge clustering dimensions.

[0053] In step 1, it is necessary to determine and use for disruptive concept innovation, specifically, the specific content of each index of the key functional features for analogical innovation knowledge, and at the same time determine the total number of functional feature indicators, and denote it as n;

[0054] Step 3 includes:

[0055] 3.1. Select the quantization criterion for the functional feature similarity of the synonym cluster to obtain each weight vector in the n functional feature sequences: w = (w1, w2,..., w n ) T ;

[0056] 3.2. Use the cosine theorem based on trigonometric function characteristics to calculate the similarity of functional features by using the word clusters of synonyms in the analogical knowledge units. The semantic similarity in the tree - type data structure of the synonym clusters is mainly determined by the depth of the concepts to be compared in the hierarchical tree, the tightness of the local area of the concepts to be compared in the hierarchical tree (the higher the tightness, the higher the similarity), the path length between the concepts to be compared (the longer the path, the lower the similarity), and the type difference of the relationships between the concepts to be compared (part - whole relationship, for part - whole relationships, the similarity between concepts of the super - sub relationship is higher). The present invention uses the cosine theorem to propose a calculation method for the similarity of functional features, specifically including the following two scenarios:

[0057] Scenario 1: Using the relationship in the synonym cluster, if two words are not on the same tree, the similarity is recorded as: Sim(A,B) = f

[0058] Scenario 2: If two words are on the same tree

[0059] If it is on the second - layer branch, the coefficient is a, and the similarity between the two is expressed by the following formula

[0060]

[0061] If it is on the third - layer branch, the coefficient is b, and the similarity between the two is expressed by the following formula

[0062]

[0063] If it is on the fourth - layer branch, the coefficient is c, and the similarity between the two is expressed by the following formula

[0064]

[0065] If it is on the fifth - layer branch, the coefficient is d, and the similarity between the two is expressed by the following formula

[0066]

[0067] In the above formula, (n - k + 1) / n is a control parameter, where n is the total number of nodes in the branch layer and k is the distance between the two branches. If the numbers of two words are the same and their word - end flag is "=", their similarity is 1; if the word - end symbol is "#", their similarity is e. After multiple experiments, the initial values of the layer numbers are set as: a = 0.65, b = 0.8, c = 0.9, d = 0.96, e = 0.5, f = 0.1.

[0068] 3.3. The similarity between eigenvectors can be obtained through the cosine theorem, thereby determining the similarity degree between concepts. The cosine theorem describes the relationship between any included angle and the three sides in a triangle. When the lengths of the three sides of a triangle are determined, the cosine theorem can be used to calculate the degree of each included angle. Suppose a triangle with side lengths a, b, and c, and the corresponding included angles A, B, and C. Then the cosine value of angle A can be calculated by the following formula.

[0069]

[0070] If sides b and c are regarded as vectors with A as the origin, the above formula can be transformed into the following formula, and the two are equivalent. In this formula, the numerator is the inner product of vectors b and c, and the geometric meaning of the denominator is the product of the lengths of the two vectors.

[0071]

[0072] Thus, by analogy, for two n-dimensional vectors such as X(x1, x2…, xn) and Y(y1, y2…, yn), the cosine of the included angle can be expressed by the following formula:

[0073]

[0074] The similarity between two concepts can be expressed as the cosine in n-dimensional Euclidean space, also known as Euclidean space, as shown in the following formula.

[0075]

[0076] In step 4, a method for quantifying technical differences is provided from the perspective of system composition. The system multi-level description model described in the analogical innovation knowledge of disruptive concept innovation can be further extended. By using the functional ontology model of the system, it is relatively easy to determine the mapping relationship between the basic functions and auxiliary functions, and then determine the implementation components or composition of the system. A complete system should include four parts: the working units, the transmission system, the energy system, and the control units.

[0077] Such as Figure 3 shown, the prototype system in the embodiment first decomposes itself into main functions and auxiliary functions according to the characteristic relationship of functions. Subsequently, whether it is the main function or the auxiliary function can be further decomposed until it can be finally decomposed to the basic function level. At the basic function level, each basic function corresponds to specific components, and these components are further combined to form the working units, transmission system, energy system, control units, etc. in the technical system.

[0078] By means of assigning weights and normalizing, the background feature vector C = [Co, Ct, Ce, Cc] for describing the functional implementation of the system is formed, corresponding to the technical difference degrees of the execution system, transmission system, power system, and control system respectively. According to the ratio of 10:6:3:1, they are used as the difference degrees of the physical principle layer, working principle layer, embodiment level, and detail level respectively, and are quantified as [1, 0.6, 0.3, 0.1] according to the normalization requirements. The knowledge nodes are compared with the system of the innovation design origin, and then the specific background feature vector values of their functional implementation are confirmed and normalized, and the specific values of the technical system difference degree are calculated and determined.

[0079] In step 5, the calculation results of the comprehensive analogy knowledge in two dimensions of functional feature similarity and technical system difference degree are integrated. As Figure 4 shown, the analogy innovation knowledge that meets the calculation threshold and has potential available value is established in a coordinate system according to two dimensions. The measured value of the functional feature is used as the abscissa of the clustering result, and the difference degree of the technical system is used as the ordinate of the clustering result. Then each knowledge unit to be clustered realizes the specific coordinates and specific positions of different analogy knowledge in the clustering system according to its functional feature parameters and the difference degree of the technical system, and completes the clustering management of the analogy innovation knowledge.

[0080] As Figure 5 shown, a method and system for clustering analogy knowledge for disruptive concept innovation of the present invention includes an operating system, application programs, an internal database, a processor, a thesaurus database, an internal memory, and a display unit. Among them, the functional modules included in the application program are as Figure 2 shown: 201 a pre-set evaluation index system; 202 a pre-set and stored module for thesaurus; 203 a functional feature similarity calculation module based on the cosine theorem; 204 a calculation method module for the technical difference degree of the technical system; 205 a clustering calculation module automatically generated according to the calculation results.

[0081] When clustering the analogy innovation knowledge, the display unit provides a relevant information input interface, the input unit receives the input data, and the application program containing five functional modules processes the received data in sequence, saves the intermediate data and analysis results of the step-by-step processing into the internal database. When the calculation and clustering are completed, the output sorting result is saved into the internal memory and displayed on the display module.

[0082] The above are only the preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A knowledge clustering method for disruptive concept innovation, characterized by: The following steps are involved: Step 1: Determine the key feature index system of functional similarity of knowledge clustering; Step 2: Conduct knowledge retrieval based on key functional indicators; Step 3: Calculate the functional similarity between clustering knowledge according to the cosine theorem method, and set the threshold according to the calculation result; Step 4: Calculate the technical difference parameters of each knowledge to be clustered; Step 5: Complete the clustering of various types of comparative knowledge through the calculation results of the two clustering indicators, namely, functional similarity and technical difference, where functional similarity and technical difference constitute two knowledge clustering dimensions.

2. The knowledge clustering method for disruptive concept innovation according to claim 1 is characterized by: In step 1, the key feature index system of functional similarity is to determine the specific content of each index of the key functional features of analog innovation knowledge, and the total number of functional feature indicators is determined, which is recorded as n.

3. The knowledge clustering method for disruptive concept innovation according to claim 1 is characterized in that: Step 3 includes: 3.

1. Select the quantitative criterion of the functional feature similarity of the synonym cluster and obtain the weight vectors of each of the n functional feature sequences: w = (w1, w2, ..., w n ) T ; 3.

2. Using the cosine theorem based on trigonometric function characteristics to calculate the similarity of functional characteristics using synonymous word clusters in analogy knowledge units, including the following two scenarios: Scenario 1: Using the relationship in the synonym cluster, if the two words are no longer in the same tree, the similarity is recorded as: Sim(A,B) = f Scenario 2: If two words are on the same tree If the branch is at the second level, the coefficient is a, and the similarity between the two is expressed by the following formula If the branch is at the third level, the coefficient is b, and the similarity between the two is expressed by the following formula If the branch is at the 4th level, the coefficient is c, and the similarity between the two is expressed by the following formula If the branch is at the 5th level, the coefficient is d, and the similarity between the two is expressed by the following formula Among them, n-k+1 / n is a control parameter, n is the total number of nodes in the branch layer, and k is the distance between two branches; if two words have the same number and their ending sign is "=", their similarity is 1, if the ending sign is "#", their similarity is e; the initial value of the number of layers is set to: a=0.65, b=0.8, c=0.9, d=0.96, e=0.5, f=0.1; 3.

3. The similarity between feature vectors is obtained by the cosine theorem to determine the similarity between concepts, specifically: The cosine expression of the angle between two n-dimensional vectors X(x1,x2…,xn) and Y(y1,y2…,yn) is as follows: The similarity between two concepts is expressed in n-dimensional Euclidean, also known as Euclidean space cosine, as shown in the following formula:

4. The knowledge clustering method for disruptive concept innovation according to claim 1 is characterized in that: Step 4 includes: 4.

1. First, confirm that the technical differences are quantified from the perspective of system composition. The components of the technical system include: execution system, transmission system, power system and control system. 4.

2. Then, by assigning weights and normalizing, a feature vector C describing the function realization background of the system is formed, C = [Co, Ct, Ce, Cc]. The four parameters in the vector correspond to the technical differences of the execution system, transmission system, power system and control system respectively. 4.

3. The differences between the physical principle layer, the working principle layer, the implementation layer and the detail layer are respectively taken as 10:6:3:1, and quantized as [1, 0.6, 0.3, 0.1] according to the normalization requirements; 4.

4. Compare the knowledge to be clustered retrieved in step 2 with the existing technical system solutions, confirm the specific function implementation background feature vector value, perform normalization, and calculate and determine the specific value of the technical system difference.

5. The knowledge clustering method for disruptive concept innovation according to claim 1 is characterized in that: Step 5 is as follows: based on the calculation results of the analogy knowledge in the two dimensions of functional feature similarity and technical system difference, a coordinate system is established according to the two dimensions for the analogy innovation knowledge that meets the calculation threshold and has potential usable value; the measurement value of the functional feature is used as the horizontal coordinate of the clustering result, and the difference of the technical system is used as the vertical coordinate of the clustering result; then each knowledge unit to be clustered realizes the specific coordinates and specific positions of different analogy knowledge in the clustering system according to its functional feature parameters and technical system difference, thereby completing the clustering of analogy innovation knowledge.

Images