Scientific and technological fusion measurement method and system and medium thereof

Abstract

The invention relates to a science and technology fusion measurement method and system and a medium thereof, and the method comprises the steps: obtaining a literature database comprising patent data and thesis data, and preliminarily constructing a science and technology knowledge network based on the literature database; perfecting the science and technology knowledge network by using a community division algorithm based on the preliminarily constructed science and technology knowledge network to obtain a science and technology knowledge community network; calculating the complementarity measure and the substitutability measure of each knowledge node in the science and technology knowledge community network; based on the complementarity measure and the substitutability measure, the fusion mode of the scientific knowledge and the technical knowledge is identified. The system and the medium are used for realizing the method. Compared with the prior art, objectivity and accuracy of science and technology fusion evaluation are improved.

Description

Technical Field

[0001] The present invention relates to the field of special data processing, and in particular to a method, system and medium for measuring technology integration. Background Art

[0002] The measurement method of technology integration is of great significance in understanding the interaction between science and technology. As technology development enters a stage of deep integration, the boundaries between science and technology are becoming increasingly blurred, cross-innovation paradigms continue to emerge, and the integration of knowledge has become a key driving force for innovation. However, current measurement methods mainly rely on traditional analytical tools, which often ignore the complex relationship structure within the science and technology knowledge network. This limitation makes it impossible to accurately capture the characteristics of knowledge integration when evaluating the level of technology integration, especially the complementary and substitutive integration characteristics of knowledge. In addition, existing methods have limited ability to identify the integration pattern of science and technology knowledge, and it is difficult to fully reflect the true face of the innovation ecosystem.

[0003] Therefore, there is an urgent need for a novel measurement framework that systematically analyzes the integration pattern of scientific and technological knowledge from the perspective of knowledge networks to support more effective science and technology innovation management and policy making. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a method, system and medium for measuring technology integration.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] According to a first aspect of the present invention, a method for measuring technology integration is provided, characterized in that the method comprises:

[0007] Acquire a literature database including patent data and paper data, and preliminarily construct a scientific and technological knowledge network based on the literature database, wherein the literature database includes the title, abstract, publication / application year field, citation relationship table and keyword table of patents and papers; the nodes of the scientific and technological knowledge network include scientific knowledge nodes and technical knowledge nodes, and the edges of the scientific and technological knowledge network are generated according to the relationship between the scientific knowledge nodes and the technical knowledge nodes;

[0008] Based on the initially constructed science and technology knowledge network, the community partitioning algorithm is used to improve the science and technology knowledge network to obtain the science and technology knowledge community network;

[0009] Calculating the complementarity measure and the substitutability measure of each of the knowledge nodes in the science and technology knowledge community network;

[0010] Identify the integration pattern of scientific knowledge and technological knowledge based on the complementarity measure and substitutability measure mentioned above.

[0011] As a preferred technical solution, the method for initially constructing a scientific and technological knowledge network is:

[0012] Designing a Prompt template of a scientific knowledge entity extraction model based on the literature database, and extracting scientific knowledge entities based on the abstract and title;

[0013] Design a prompt template of a technical knowledge entity extraction model based on the scientific knowledge entity, and extract the technical knowledge entity based on the patent data;

[0014] Cleaning and organizing the scientific knowledge entities and technical knowledge entities;

[0015] The cleaned and organized scientific knowledge entities and technical knowledge entities are used as nodes of the scientific and technological knowledge network, and the co-occurrence relationship between the two entities is extracted as the edge of the scientific and technological knowledge network.

[0016] As a preferred technical solution, the structures of the Prompt template of the scientific knowledge entity extraction model and the Prompt template of the technical knowledge entity extraction model are both:

[0017] Prompt=Prefix+Instruct+Examples+Input,

[0018] Among them, Prefix is ​​used to set the role of the entity extraction model; Instruct is used for task description; Examples represents task examples. If no task examples are set, it indicates that the task is zero-sample learning; Input represents the summary and title data to be processed.

[0019] As a preferred technical solution, the community division algorithm performs the following steps:

[0020] Dividing the scientific and technological knowledge network into a plurality of initial communities, and each initial community has only one node;

[0021] Select multiple different starting mobile nodes, and perform the three steps of node movement, new network construction, and community structure acquisition to obtain multiple community structures:

[0022] Node movement: Calculate the modularity gain of moving each node to the adjacent node community, select the target community for node movement according to the preset rules based on the modularity gain, and execute the movement to obtain a new community, repeating this step until there is no modularity gain for node movement;

[0023] New network construction: Take each new community as a super node, construct the edges between the super nodes, calculate the edge weights between the super nodes, and generate a new network;

[0024] Acquiring community structure: Repeat the steps of node movement and new network construction until the network has no positive modularity gain and obtains the community structure;

[0025] The voting mechanism is used to count the number of times the same node appears in the same community in different community structures, and a similarity matrix is ​​constructed based on the number of times.

[0026] An intermediate network is constructed based on the similarity matrix, and the three steps of node movement, new network construction and community structure acquisition are performed again based on the intermediate network to obtain a scientific and technological knowledge community network.

[0027] As a preferred technical solution, the preset rules are:

[0028] Calculate the modularity gain of the node moving to the adjacent community. If only one adjacent community has a positive modularity gain, the adjacent community corresponding to the positive gain is used as the target moving community.

[0029] If there are multiple positive modularity gains, the adjacent community corresponding to the maximum positive gain is selected as the target mobile community;

[0030] If the maximum positive gains are the same, the community with fewer nodes in the adjacent community corresponding to the maximum positive gain is selected as the target mobile community;

[0031] If the maximum positive gains are the same and the communities corresponding to the maximum positive gains contain the same number of nodes, a community is randomly selected as the target mobile community.

[0032] As a preferred technical solution, the method for calculating the complementarity measure is:

[0033] Get the preset node co-occurrence count and the average of the node co-occurrence frequency, and calculate the first intermediate value according to the following formula:

[0034]

[0035] Among them, C jk represents the number of co-occurrences of node j and node k in the science and technology knowledge community network; μ jk represents the mean of the co-occurrence frequency between nodes j and k in the network under the random assumption, and the mean is approximately estimated based on the mean of the co-occurrence frequency of all known two nodes in the science and technology knowledge community network; σ jk represents the standard deviation; jk represents the first intermediate value;

[0036] The first intermediate value is standardized, and the expression is:

[0037]

[0038] Among them, Minλ jk Indicates the minimum value of the first intermediate value, Maxλ jk represents the maximum value of the first intermediate value, λ′ jk represents the second intermediate value;

[0039] The total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node is obtained as the weight, and the third intermediate value is calculated based on the weight. The expression is:

[0040]

[0041] Among them, WAR j represents the third intermediate value, j and k both represent nodes, P k Represents the corresponding weight value of knowledge node k;

[0042] The complementarity measure is calculated based on the third intermediate value, and its expression is:

[0043]

[0044] Among them, Λ1 represents the complementarity measure, P j represents the weight value corresponding to knowledge node j, and n represents the total number of nodes.

[0045] As a preferred technical solution, the method for calculating the alternative measure is:

[0046] Calculate the fourth median value using cosine similarity Its expression is:

[0047]

[0048] Among them, C jm represents the number of co-occurrences of nodes j and m in the science and technology knowledge community network; C km represents the number of co-occurrences of node k and node m; n represents the total number of nodes;

[0049] The total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node is obtained as the weight, and the fifth intermediate value is calculated based on the weight. The expression is:

[0050]

[0051] in, represents the fourth intermediate value, P k Represents the corresponding weight value of knowledge node k, WAR′ j represents the fifth median value;

[0052] An alternative measure is calculated based on the fifth intermediate value, and its expression is:

[0053]

[0054] Among them, Λ2 represents the alternative measure, P j Represents the weight value corresponding to knowledge node j.

[0055] As a preferred technical solution, the method for identifying the fusion pattern of scientific knowledge and technical knowledge is:

[0056] The complementarity measure and the substitution measure are set as the observed variables of the hidden Markov model, and the fusion mode of scientific knowledge and technological knowledge is set as the hidden state of the hidden Markov model;

[0057] estimating model parameters of a hidden Markov model, wherein the model parameters include initial state probability, state transition probability, and distribution of observed variables under different states;

[0058] Based on the estimated model parameters, the Viterbi algorithm is used to decode the observed variables in different time windows and identify the hidden states;

[0059] Based on the complementarity measure and the substitution measure, the average complementarity measure and the average substitution measure of the knowledge of each hidden state are calculated, and based on the average complementarity measure and the average substitution measure, a fusion mode is obtained, and the fusion mode includes fusion, complementation and substitution.

[0060] As a preferred technical solution, the method for estimating the model parameters of the hidden Markov model is:

[0061] The forward-backward algorithm is used to estimate the model parameters, and the expression is:

[0062]

[0063] Among them, λ MLE represents the maximum likelihood estimate of the model parameter λ, P(O|λ) represents the probability of the observation sequence O occurring, It means to find the value of λ so that P(O|λ) or ∑ S logP(O,S|λ)P(S|O,λ t ) reaches the maximum parameter value; S represents the hidden state; P(S|O,λ t ) means that given the observation sequence O and the current model parameter λ t The posterior probability of the state sequence S under the given model parameter λ; P(O,S|λ) represents the joint logarithmic probability of the observation sequence O and the state sequence S under the given model parameter λ.

[0064] According to a second aspect of the present invention, there is provided a system for measuring technology integration, comprising a memory and a processor, wherein a computer program is stored in the memory, and the above method is implemented when the processor executes the program.

[0065] Compared with the prior art, the present invention has the following advantages:

[0066] 1) This application provides a quantitative analysis method for the integration of scientific knowledge and technological knowledge. By calculating the substitution and complementarity measures of knowledge, the degree of interaction between science and technology is objectively evaluated. Compared with the traditional subjective evaluation method, the method provided by this application is based on data-driven statistical analysis, which reduces the bias of human evaluation and improves the objectivity and accuracy of the integration measurement.

[0067] 2) This application effectively combines the community partitioning algorithm with the hidden Markov model, which can process large-scale data and divide the constructed knowledge network into multiple communities. Through the interaction between community detection and knowledge nodes, it can quickly capture the complex relationship and fusion characteristics between science and technology, thereby improving computing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is a flow chart of the method of the present invention;

[0069] Figure 2 Visual analysis of the scientific and technological knowledge network of the present invention; (2a) the scientific and technological knowledge network before 2000; (2b) the scientific and technological knowledge network from 2000 to 2016; (2c) the scientific and technological knowledge network from 2016 to the present. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0071] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0072] Example 1

[0073] This embodiment provides a method for measuring the knowledge fusion between science and technology, extracting features from scientific papers and technical patents to accurately reflect the internal mechanism of scientific and technological innovation. The process of this method is as follows: Figure 1 The detailed steps are as follows:

[0074] S1. Initially build a scientific and technological knowledge network.

[0075] S11. Download a literature database including patent data and paper data, and classify, merge and clean the data in the literature database, wherein the literature database includes the title, abstract, publication / application year field, citation relationship table and keyword table of patents and papers.

[0076] S12. Design a prompt template for the scientific knowledge entity extraction model based on the processed literature database, and extract scientific knowledge entities based on the abstract and title. The design principle of the prompt template is to minimize the hallucination content generated by the large model while ensuring that the results are accurate and easy to interpret. Its design follows the following structure:

[0077] Prompt=Prefix+Instruct+Examples+Input,

[0078] Among them, Prefix is ​​used to set the role of the entity extraction model; Instruct is used for task description; Examples represents task examples. If no task examples are set, it indicates that the task is zero-sample learning; Input represents the summary and title data to be processed.

[0079] The extraction template of scientific knowledge entities designed according to the above structure is shown in Table 1.

[0080] S13. Design a Prompt template for the technical knowledge entity extraction model based on scientific knowledge entities. The structure of the Prompt template is consistent with the Prompt template structure for extracting scientific knowledge entities. Technical knowledge entities are extracted based on patent data. The template for extracting technical knowledge entities is shown in Table 1.

[0081] Table 1 Prompt prompt example table

[0082]

[0083]

[0084] S14. Clean and organize scientific knowledge entities and technical knowledge entities.

[0085] The keywords extracted during the extraction of scientific knowledge entities and technical knowledge entities are cleaned, including deleting incorrectly extracted sentences and keywords of non-knowledge entities.

[0086] According to steps S11 to S14, knowledge entities were extracted from the literature database including 9720 scientific papers and technical patents. The details of the extraction are shown in Table 2. It can be seen that a total of 22569 technical knowledge entities were extracted, with a total frequency of 70460 times, and an average of 8.12 technical keywords per document; in contrast, 4287 scientific knowledge entities were extracted, with a total frequency of 31595 times, and an average of 3.64 scientific keywords per document. That is, the number and total frequency of technical knowledge entities are significantly higher than those of scientific knowledge entities, reflecting the diversity and complexity of the field of science and technology, involving richer concepts and terms.

[0087] Table 2 Extraction result information table

[0088]

[0089] S15. The cleaned and organized scientific knowledge entities and technical knowledge entities are used as nodes of the scientific and technological knowledge network, and the co-occurrence relationship between the two entities is extracted as the edge of the scientific and technological knowledge network.

[0090] Taking the field of artificial intelligence as an example, the research time is divided into three stages according to its development history, namely before 2000, from 2000 to 2016, and from 2016 to the present. The corresponding scientific and technological knowledge networks are constructed according to the divided event segments, such as Figure 2 As shown, Figure (2a) is the science and technology knowledge network before 2000, Figure (2b) is the science and technology knowledge network from 2000 to 2016, and Figure (2c) is the science and technology knowledge network from 2016 to the present.

[0091] It can be seen that before 2000, the scale of the science and technology knowledge network was small, and the combination relationship between technical knowledge entities (red nodes) and scientific knowledge entities (blue nodes) was relatively scattered. There were only 4666 nodes and 23136 edges in the network, indicating that the degree of integration of technical knowledge with scientific knowledge was limited during this period. However, an obvious cluster appeared in the network, indicating that a group of technical knowledge was integrated with a certain scientific knowledge, which may represent the application exploration of early technical knowledge in a specific field. From 2000 to 2016, the scale of the science and technology knowledge network expanded significantly, with the number of nodes increasing to 11990 and the number of edges increasing to 69366. Several obvious technical knowledge clusters appeared, indicating that different The technological knowledge of the 20th century has been applied in more scientific knowledge fields. The network structure of this period reflects the diversified application of technological knowledge in different industries. The number of nodes and edges has increased, reflecting the closer integration relationship between technology and science. From 2016 to the present, the scale of the science and technology knowledge network has further expanded, with the number of nodes reaching 15,667 and the number of edges increasing to 100,920. The density and connectivity of the network have significantly improved, reflecting that the penetration of technological knowledge in scientific knowledge has been greatly enhanced, promoting the intelligent and digital transformation of scientific knowledge. Dense clusters of scientific knowledge entities have appeared in the network, indicating that some common technological knowledge has been widely used in multiple scientific knowledge fields, forming a large number of knowledge integration relationships. By drawing the network diagrams of each stage, the changes of the network structure over time are intuitively displayed. From the characteristic indicators of the science and technology knowledge network of each period summarized in Table 3, it can be seen that with the passage of time, the number of nodes and edges has increased significantly. The increase in average degree shows that each entity has formed a fusion relationship with more other entities, and the correlation between nodes has increased. These analyses provide important empirical evidence for understanding the deep integration of technological knowledge and scientific knowledge.

[0092] Table 3. Sample information table in the field of artificial intelligence

[0093]

[0094] S2. Based on the science and technology knowledge network constructed in step S1, the community partitioning algorithm is used to improve the science and technology knowledge network to obtain the science and technology knowledge community network.

[0095] S21 and Louvain algorithms are efficient and adaptable and can handle large-scale networks. Therefore, they are selected as the main community division tools to divide the scientific and technological knowledge network into multiple initial communities, and each initial community has only one node.

[0096] S22. In order to improve the stability of the community division results, the Louvain algorithm is executed multiple times. The Louvain algorithm selects multiple different starting mobile nodes, and executes the three steps of node movement, new network construction and community structure acquisition respectively to obtain multiple community structures:

[0097] S221, Node Movement:

[0098] The modularity gain of moving each node to the adjacent node community is calculated, and the target community for node movement is selected according to the preset rules based on the modularity gain and the movement is performed to obtain a new community. Step S221 is repeated until there is no positive modularity gain for the node movement.

[0099] Specifically, the preset rule for node movement is: calculate the modularity gain of the node moving to the adjacent community. If only one adjacent community has a positive modularity gain, the adjacent community corresponding to the positive gain is used as the target moving community; if multiple modularity gains are positive, the adjacent community corresponding to the maximum positive gain is selected as the target moving community; if the maximum positive gains are the same, the community with fewer nodes in the adjacent community corresponding to the maximum positive gain is selected as the target moving community; if the maximum positive gains are the same and the communities corresponding to the maximum positive gains contain the same number of nodes, a community is randomly selected as the target moving community.

[0100] S222, New network construction:

[0101] Each new community is used as a super node, the edges between the super nodes are constructed, the edge weights between the super nodes are calculated, and a new network is generated.

[0102] Specifically, the edge weight between super nodes is the total edge weight between corresponding communities in the original network, that is, considering all cross-community connections. For example, the edge weight between community A and community B is the sum of the weights of all edges from any node in A to any node in B in the original network.

[0103] S223. Community structure acquisition:

[0104] The steps of node movement and new network construction are repeated continuously until the network has no positive modularity gain and a community structure is obtained.

[0105] S23. Use a voting mechanism to count the number of times the same node appears in the same community in different community structures, and build a similarity matrix based on the number. The elements of the matrix represent the probability that two nodes are in the same community. The higher the probability, the more likely it is that the two nodes belong to the same community.

[0106] S24. Construct an intermediate network based on the similarity matrix, and perform the three steps of node movement, new network construction and community structure acquisition again based on the intermediate network to obtain a scientific and technological knowledge community network.

[0107] The science and technology knowledge community network includes the following eight communities. The detailed community information is shown in Table 4:

[0108] Table 4 Community information table

[0109] Community ID Community size Number of edges Average Maximum minimum degree 1 5290 27267 10.51359 773 1 2 8001 35226 8.954245 2379 1 3 4260 41509 19.82283 1124 1 4 4184 19274 9.480571 867 1 5 4773 22898 9.840138 1950 1 6 362 1323 7.538462 119 1 7 7 12 3.428571 6 1 8 3 2 1.333333 2 1

[0110] As can be seen from Table 4, the community size varies from 3 nodes to 8001 nodes, and the number of edges also varies accordingly. The largest number of edges is the community numbered 3, which has 41,509 edges; the smallest number of edges is the community numbered 8, which has 2 edges. In addition, the average degree indicates the degree of integration of each node with other nodes. The community numbered 3 has the highest average degree, which is 19.82, indicating that the nodes in this community have the strongest correlation.

[0111] S3. Compute the complementarity measure and substitutability measure of each knowledge node in the science and technology knowledge community network.

[0112] For each community's scientific knowledge entity and technical knowledge entity in the scientific and technological knowledge community network, their complementarity and substitutability are calculated to accurately evaluate the knowledge integration characteristics between science and technology. Specifically, the complementarity measure evaluates the synergy between two knowledge entities, while the substitutability measure measures the degree of overlap between two entities in the same field. The detailed steps are as follows:

[0113] S31, obtaining a preset number of node co-occurrences and an average of the node co-occurrence frequencies, and calculating a first intermediate value according to the following formula:

[0114]

[0115] Among them, C jk represents the number of co-occurrences of node j and node k in the science and technology knowledge community network; μ jk represents the mean value of the co-occurrence frequency between nodes j and k in the network under the random assumption, and this value is approximately estimated based on the mean value of the co-occurrence frequency of all known two nodes in the science and technology knowledge community network; σ jk represents the standard deviation; jkIndicates the first intermediate value.

[0116] S32, standardize the first intermediate value, and the expression is:

[0117]

[0118] Minλ jk Indicates the minimum value of the first intermediate value, Maxλ jk represents the maximum value of the first intermediate value, λ′ jk Indicates the second intermediate value.

[0119] S33. The total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node is obtained as a weight, and the third intermediate value is calculated based on the weight, and the expression is:

[0120]

[0121] WAR j represents the third intermediate value, j and k both represent nodes, P k Represents the corresponding weight value of knowledge node k.

[0122] S34. Calculate the complementarity measure based on the third intermediate value, and the expression thereof is:

[0123]

[0124] Λ1 represents the complementarity measure, P j represents the weight value corresponding to knowledge node j, and n represents the total number of nodes.

[0125] S35, calculating the fourth intermediate value using cosine similarity Its expression is:

[0126]

[0127] C jm represents the number of co-occurrences of nodes j and m in the science and technology knowledge community network; C km represents the number of co-occurrences of node k and node m; n represents the total number of nodes.

[0128] S36. Obtain the total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node as a weight, and calculate the fifth intermediate value based on the weight, which is expressed as follows:

[0129]

[0130] represents the fourth intermediate value, P k Represents the corresponding weight value of knowledge node k, WAR′ j Indicates the fifth intermediate value.

[0131] S37. Calculate an alternative measure based on the fifth intermediate value, the expression of which is:

[0132]

[0133] Λ2 represents the alternative measure, P j Represents the weight value corresponding to knowledge node j.

[0134] The calculation is performed using communities numbered 1 and 2 as examples. In community numbered 1, the co-occurrence relationship between scientific knowledge entities and technical knowledge entities is relatively dense, and the complementarity measure value in this community is the highest, mainly concentrated in the fields of intelligent manufacturing and big data, reflecting the high intensity of scientific and technological integration in these fields. For example, there is a high degree of synergy between technical knowledge entities in intelligent manufacturing and scientific knowledge entities in big data analysis, and the complementarity measure value λ is jk It reaches 0.85, showing that the knowledge integration in these fields has a strong driving force for innovation-driven development; in the community numbered 2, the substitution measurement value of scientific knowledge entities and technical knowledge entities is high, especially in the fields of automation control technology and mechanical engineering. Reaching 0.75 indicates that in these fields, the application of technical knowledge may partially replace certain contents of scientific research. For example, the application of automatic control technology in mechanical engineering has gradually replaced some traditional scientific experimental methods and promoted industrial upgrading.

[0135] S4. Identify the integration pattern of scientific knowledge and technological knowledge based on complementarity measures and substitutability measures.

[0136] S41. Set the complementarity measure and the substitutability measure as the observed variables of the hidden Markov model, and set the fusion mode of scientific knowledge and technological knowledge as the hidden state of the hidden Markov model.

[0137] S42. Estimate model parameters of the hidden Markov model, wherein the model parameters include initial state probability, state transition probability, and distribution of observed variables under different states.

[0138] Step S42 also includes calculating the Akaike Information Criterion (AIC) value and the Bayesian Information Criterion (BIC) value based on the estimated hidden Markov model parameters. Both of these indicators help select the optimal model configuration by comparing the model performance under different numbers of hidden states to ensure the most appropriate knowledge fusion mode. The smaller the AIC value, the better the model, and the smaller the BIC value, the better the model. The detailed calculation method is:

[0139] AIC = 2k-2ln(L),

[0140] BIC = ln(n)k-2ln(L),

[0141] Among them, k is the number of free parameters in the model, L is the maximum likelihood estimate of the model, and n is the number of samples.

[0142] S421, using the forward-backward algorithm to estimate the model parameter λ=(π, A, B), the expression of which is:

[0143]

[0144] Among them, λ MLE represents the maximum likelihood estimate of the model parameter λ, P(O|λ) represents the probability of the observation sequence O occurring, It means to find the value of λ so that P(O|λ) or ∑ s logP(O,S|λ)P(S|O,λ t ) reaches the maximum parameter value; S represents the hidden state; P(S|O,λ t ) means that given the observation sequence O and the current model parameter λ t P(O, S|λ) represents the joint logarithmic probability of the observation sequence O and the state sequence S under the given model parameter λ, π is the initial probability of each state, A is the transition probability between states, and B is the distribution of the observed variable O under different states.

[0145] S43. Based on the estimated model parameters, the Viterbi algorithm is used to decode the observed variables in different time windows to identify the hidden states.

[0146] S44. Based on the complementarity measure and the substitution measure, the average complementarity measure and the average substitution measure of the knowledge of each hidden state are calculated, and a fusion mode is obtained based on the average complementarity measure and the average substitution measure, and the fusion mode includes fusion, complementation and substitution.

[0147] Specifically, when the average complementarity measure and the average substitution measure of community knowledge in the science and technology knowledge community network are both greater than 0.1, and the difference between the two is less than 0.2, the corresponding fusion mode is defined as "fusion"; if the average complementarity measure is greater than or equal to the average substitution measure, the corresponding fusion mode is defined as "complementarity"; if the average complementarity measure is less than the average substitution measure, the corresponding fusion mode is defined as "substitution". More specifically, when the fusion mode is "complementarity" and "substitution", it can be divided into "low (0-0.2)", "medium (0.2-0.4)" and "high (greater than 0.4)" according to the average complementarity measure or the average substitution measure, respectively.

[0148] In this embodiment, six major communities in the science and technology knowledge community network are analyzed, and five hidden states are identified. The detailed information is shown in Table 5:

[0149] Table 5 Knowledge fusion mode table

[0150]

[0151] According to Table 5, in the context of state 1, the integration of scientific and technological knowledge is most significant, while state 5 reflects strong knowledge substitutability. This analysis provides a quantitative basis for understanding the integration model of scientific and technological knowledge and can guide the formulation of further science and technology development strategies.

[0152] Example 2

[0153] The above is an introduction to the method embodiment. The following is a further explanation of the scheme of the present invention through a system embodiment. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process described can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0154] The present embodiment provides a system for technology integration measurement, which includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or computer program instructions loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0155] Multiple components in the device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.

[0156] The processing unit performs the various methods and processes described above, such as methods S1 to S4. For example, in some embodiments, methods S1 to S4 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S4 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S4 in any other appropriate manner (e.g., by means of firmware).

[0157] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0158] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0159] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0160] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for measuring technology integration, characterized in that: The method includes: Acquire a literature database including patent data and paper data, and preliminarily construct a scientific and technological knowledge network based on the literature database, wherein the literature database includes the title, abstract, publication / application year field, citation relationship table and keyword table of patents and papers; the nodes of the scientific and technological knowledge network include scientific knowledge nodes and technical knowledge nodes, and the edges of the scientific and technological knowledge network are generated according to the relationship between the scientific knowledge nodes and the technical knowledge nodes; Based on the initially constructed science and technology knowledge network, the community partitioning algorithm is used to improve the science and technology knowledge network to obtain the science and technology knowledge community network; Calculating the complementarity measure and the substitutability measure of each of the knowledge nodes in the science and technology knowledge community network; Identify the integration pattern of scientific knowledge and technological knowledge based on the complementarity measure and substitutability measure mentioned above.

2. A method for measuring technology integration according to claim 1, characterized in that: The method for initially constructing a scientific and technological knowledge network is as follows: Designing a Prompt template of a scientific knowledge entity extraction model based on the literature database, and extracting scientific knowledge entities based on the abstract and title; Design a prompt template of a technical knowledge entity extraction model based on the scientific knowledge entity, and extract the technical knowledge entity based on the patent data; Cleaning and organizing the scientific knowledge entities and technical knowledge entities; The cleaned and organized scientific knowledge entities and technical knowledge entities are used as nodes of the scientific and technological knowledge network, and the co-occurrence relationship between the two entities is extracted as the edge of the scientific and technological knowledge network.

3. A method for measuring technology integration according to claim 2, characterized in that: The structures of the Prompt template of the scientific knowledge entity extraction model and the Prompt template of the technical knowledge entity extraction model are both: Prompt=Prefix+Instruct+Examples+Input, Among them, Prefix is ​​used to set the role of the entity extraction model; Instruct is used for task description; Examples represents task examples. If no task examples are set, it indicates that the task is zero-sample learning; Input represents the summary and title data to be processed.

4. A method for measuring technology integration according to claim 1, characterized in that: The community decomposition algorithm performs the following steps: Dividing the scientific and technological knowledge network into a plurality of initial communities, and each initial community has only one node; Select multiple different starting mobile nodes, and perform the three steps of node movement, new network construction, and community structure acquisition to obtain multiple community structures: Node movement: Calculate the modularity gain of moving each node to the adjacent node community, select the target community for node movement according to the preset rules based on the modularity gain, and execute the movement to obtain a new community, repeating this step until there is no modularity gain for node movement; New network construction: Take each new community as a super node, construct the edges between the super nodes, calculate the edge weights between the super nodes, and generate a new network; Acquiring community structure: Repeat the steps of node movement and new network construction until the network has no positive modularity gain and obtains the community structure; The voting mechanism is used to count the number of times the same node appears in the same community in different community structures, and a similarity matrix is ​​constructed based on the number of times. An intermediate network is constructed based on the similarity matrix, and the three steps of node movement, new network construction and community structure acquisition are performed again based on the intermediate network to obtain a scientific and technological knowledge community network.

5. A method for measuring technology integration according to claim 4, characterized in that: The preset rules are: Calculate the modularity gain of the node moving to the adjacent community. If only one adjacent community has a positive modularity gain, the adjacent community corresponding to the positive gain is used as the target moving community. If there are multiple positive modularity gains, the adjacent community corresponding to the maximum positive gain is selected as the target mobile community; If the maximum positive gains are the same, the community with fewer nodes in the adjacent community corresponding to the maximum positive gain is selected as the target mobile community; If the maximum positive gains are the same and the communities corresponding to the maximum positive gains contain the same number of nodes, a community is randomly selected as the target mobile community.

6. A method for measuring technology integration according to claim 1, characterized in that: The method for calculating the complementarity measure is: Get the preset node co-occurrence count and the average of the node co-occurrence frequency, and calculate the first intermediate value according to the following formula: Among them, C jk represents the number of co-occurrences of node j and node k in the science and technology knowledge community network; μ jk represents the mean of the co-occurrence frequency between nodes j and k in the network under the random assumption, and the mean is approximately estimated based on the mean of the co-occurrence frequency of all known two nodes in the science and technology knowledge community network; σ jk represents the standard deviation; jk represents the first intermediate value; The first intermediate value is standardized, and the expression is: Among them, Minλ jk Indicates the minimum value of the first intermediate value, Maxλ jk represents the maximum value of the first intermediate value, λ j ′ k represents the second intermediate value; The total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node is obtained as the weight, and the third intermediate value is calculated based on the weight. The expression is: Among them, WAR j represents the third intermediate value, j and k both represent nodes, P k Represents the corresponding weight value of knowledge node k; The complementarity measure is calculated based on the third intermediate value, and its expression is: Among them, Λ1 represents the complementarity measure, P j represents the weight value corresponding to knowledge node j, and n represents the total number of nodes.

7. A method for measuring technology integration according to claim 6, characterized in that: The method for calculating the alternative measure is: Calculate the fourth median value using cosine similarity Its expression is: Among them, C jm represents the number of co-occurrences of nodes j and m in the science and technology knowledge community network; C km represents the number of co-occurrences of node k and node m; n represents the total number of nodes; The total number of patents and papers that overlap the scientific knowledge node and the technical knowledge node is obtained as the weight, and the fifth intermediate value is calculated based on the weight. The expression is: in, represents the fourth intermediate value, P k Represents the corresponding weight value of knowledge node k, WAR′ j represents the fifth median value; An alternative measure is calculated based on the fifth intermediate value, and its expression is: Among them, Λ2 represents the alternative measure, P j Represents the weight value corresponding to knowledge node j.

8. A method for measuring technology integration according to claim 1, characterized in that: The method for identifying the integration pattern of scientific knowledge and technical knowledge is: The complementarity measure and the substitution measure are set as the observed variables of the hidden Markov model, and the fusion mode of scientific knowledge and technological knowledge is set as the hidden state of the hidden Markov model; estimating model parameters of a hidden Markov model, wherein the model parameters include initial state probability, state transition probability, and distribution of observed variables under different states; Based on the estimated model parameters, the Viterbi algorithm is used to decode the observed variables in different time windows and identify the hidden states; Based on the complementarity measure and the substitution measure, the average complementarity measure and the average substitution measure of the knowledge of each hidden state are calculated, and based on the average complementarity measure and the average substitution measure, a fusion mode is obtained, and the fusion mode includes fusion, complementation and substitution.

9. A method for measuring technology integration according to claim 8, characterized in that: The method for estimating the model parameters of the hidden Markov model is: The forward-backward algorithm is used to estimate the model parameters, and the expression is: Among them, λ MLE represents the maximum likelihood estimate of the model parameter λ, P(O|λ) represents the probability of the observation sequence O occurring, Denotes the search for λ such that P(O|λ) or ∑ S logP(O,S|λ)P(S|O,λ t ) reaches the maximum parameter value; S represents the hidden state; P(S|O,λ t ) means that given the observation sequence O and the current model parameter λ t The posterior probability of the state sequence S under the given model parameter λ; P(O,S|λ) represents the joint logarithmic probability of the observation sequence O and the state sequence S under the given model parameter λ.

10. A system for measuring technology integration, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 9 is implemented.

Images