Devices and methods for providing information about fusion technologies and new items

Abstract

An apparatus and method for providing information on convergence technology and new items are disclosed. A keyword extraction unit extracts a plurality of patent topics from patent disclosure information and extracts a plurality of trademark topics from trademark disclosure information; and extracts similarity between a plurality of technology keywords included in each of the patent topics and a plurality of product keywords included in each of the trademark topics, and calculates the patent topics and trademark topics with high similarity.

Description

Technology Field

[0001] The present invention relates to an apparatus and method for providing information on convergence technology and new items, and more specifically, to an apparatus and method for providing information on convergence technology and new items capable of visualizing similarity between technology, products, convergence technology information, and new item information. Background Technology

[0002] Recently, technologies across various fields are evolving in diverse forms through the advancement of single technologies or linkages with others. In particular, there is a distinctly increasing trend of two or more technology groups interacting to emerge as convergence technologies. Furthermore, many institutions and researchers are actively conducting analyses to predict promising future convergence technologies.

[0003] Convergence technology promotes technological advancement by collaborating and integrating progress across various technological fields, and contributes to the development and innovation of new items through novel ideas and solutions. These new items play a crucial role in enhancing market competitiveness and creating new value.

[0004] Generally, research aimed at identifying promising items is conducted by leading domestic and international research institutions using industry-level and qualitative methodologies for the purpose of selecting promising national industries or technology areas, and there is no system applicable to the business level conducted by companies.

[0005] Although a device and method for providing information on promising items have been disclosed in relation to this, they have limitations as a unidirectional system for providing information on promising items, and consequently, it is difficult to identify the interactions between technology, products, convergence technology, and new items. The problem to be solved

[0006] The present invention provides a device and method for providing convergence technology and new item information capable of visualizing similarity between technology, product, convergence technology information, and new item information. means of solving the problem

[0007] The apparatus and method for providing convergence technology and novel item information according to the present invention comprises: a keyword extraction unit that extracts a plurality of patent topics from patent disclosure information and extracts a plurality of trademark topics from trademark disclosure information; and a method for extracting similarity between a plurality of technology keywords included in each of the patent topics and a plurality of product keywords included in each of the trademark topics, and for calculating the patent topics and trademark topics with high similarity.

[0008] In addition, the keyword extraction unit can extract the patent topic and the trademark topic using any one of TopSBM (Topic Models based on Stochastic Block Models), Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), Non-Negative Matrix Factorization (NMF), Hierarchical Dirichlet Process (HDP), and Biterm Topic Model (BTM).

[0009] In addition, the similarity measurement unit can extract the similarity based on Kullback-Leibler divergence (KL-Divergence).

[0010] In addition, the keyword extraction unit can extract IPC keywords from IPC information, and the similarity measurement unit can extract the similarity between the IPC keywords and the technology keywords.

[0011] In addition, the keyword extraction unit can extract industry keywords from the classification code and description of KSIC, and the similarity measurement unit can extract the similarity between the industry keywords and the product keywords.

[0012] Additionally, the method may include a keyword extraction step of extracting multiple patent topics from patent disclosure information and extracting multiple trademark topics from trademark disclosure information; and a similarity measurement step of extracting similarity between multiple technical keywords included in each of the patent topics and multiple product keywords included in each of the trademark topics, and calculating the patent topics and trademark topics with high similarity.

[0013] In addition, the keyword extraction step can extract the patent topic and the trademark topic using any one of TopSBM (Topic Models based on Stochastic Block Models), Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), Non-Negative Matrix Factorization (NMF), Hierarchical Dirichlet Process (HDP), and Biterm Topic Model (BTM).

[0014] In addition, the similarity measurement step can extract the similarity based on Kullback-Leibler divergence (KL-Divergence).

[0015] In addition, the keyword extraction step extracts IPC keywords from IPC information, and the similarity measurement step can extract the similarity between the IPC keywords and the technology keywords.

[0016] In addition, the keyword extraction step can extract industry keywords from the classification code and description of KSIC, and the similarity measurement step can extract the similarity between the industry keywords and the product keywords. Effects of the invention

[0017] According to the present invention, information on new fusion technologies and new items can be visualized and provided.

[0018] In addition, according to the present invention, technology and item information for discovering commercialization opportunities can be provided. Brief explanation of the drawing

[0019] FIG. 1 is a drawing showing a device for providing information on convergence technology and new items according to an embodiment of the present invention. FIG. 2 is a diagram illustrating the process of a device providing information on fusion technology and new item information according to an embodiment of the present invention. Figure 3 is a diagram showing seven patent topics extracted from the keyword extraction unit of Figure 2 and patent keywords included in each patent topic. Figure 4 is a diagram showing 23 trademark topics extracted from the keyword extraction unit of Figure 2 and trademark keywords included in each trademark topic. Figure 5 shows the probability that the title of the invention 'weather user interface' of the 531st patent disclosure information among the 1,790 patent disclosure information of Figure 2 is included in each patent topic. Figure 6 is a diagram showing the probability that the first trademark keyword 'IPHONE 11 (Trademark name)' among the 140 trademark information items of Figure 2 is included in each trademark topic by the method described above. FIG. 7 is a drawing illustrating a method for calculating similarity between technology and products according to an embodiment of the present invention. Figure 8 is a diagram showing the result value calculated by the method of Figure 7. FIG. 9 is a drawing illustrating a method for calculating similarity between technology and products according to another embodiment of the present invention. Figure 10 is a diagram showing the result value calculated by the method of Figure 9. Figure 11 is a drawing displaying the similarity between the technology and product of Figure 8. FIG. 12 is a diagram showing fusion technology information displayed in a display step according to an embodiment of the present invention. FIG. 13 is a drawing showing new item information displayed in a display step according to an embodiment of the present invention. Specific details for implementing the invention

[0020] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. However, the technical concept of the present invention is not limited to the embodiments described herein and may be embodied in other forms. Rather, the embodiments introduced herein are provided to ensure that the disclosed content is thorough and complete and to sufficiently convey the concept of the present invention to those skilled in the art.

[0021] In this specification, when a component is described as being on another component, it means that it may be formed directly on the other component or that a third component may be interposed between them. Additionally, in the drawings, the thicknesses of the films and regions are exaggerated for the effective description of the technical content.

[0022] Additionally, although terms such as first, second, third, etc., have been used to describe various components in the various embodiments of this specification, these components should not be limited by such terms. These terms are used merely to distinguish one component from another. Accordingly, what is referred to as the first component in one embodiment may be referred to as the second component in another embodiment. Each embodiment described and illustrated herein also includes its complementary embodiment. Furthermore, in this specification, "and / or" is used to mean including at least one of the components listed before and after it.

[0023] In the specification, singular expressions include plural expressions unless the context clearly indicates otherwise. Furthermore, terms such as "include" or "have" are intended to specify the existence of the features, numbers, steps, components, or combinations thereof described in the specification, and should not be understood as excluding the existence or addition of one or more other features, numbers, steps, components, or combinations thereof. Additionally, in this specification, "connection" is used to include both indirectly connecting multiple components and directly connecting them.

[0024] In addition, in describing the present invention below, if it is determined that a detailed description of related known functions or configurations could unnecessarily obscure the essence of the invention, such detailed description will be omitted.

[0026] FIG. 1 is a drawing showing a device (10) providing information on fusion technology and new items according to an embodiment of the present invention.

[0027] Referring to FIG. 1, the device (10) for providing information on convergence technology and new items includes a keyword extraction unit (100), a similarity measurement unit (200), and a display unit (300).

[0028] The keyword extraction unit (100) extracts keywords from text data. The keyword extraction unit (100) can extract only nouns from the text data, remove particles, special characters, symbols, numbers, etc., and infer what keywords exist. Specifically, the keyword extraction unit (100) analyzes the patterns of words occurring within the text data, groups words with similar meanings to extract keywords, and probabilistically estimates what words each keyword contains. The keyword extraction unit (100) can extract keywords using any one of TopSBM (Topic Models based on Stochastic Block Models), Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), Non-Negative Matrix Factorization (NMF), Hierarchical Dirichlet Process (HDP), and Biterm Topic Model (BTM).

[0029] TopSBM (Topic Models based on Stochastic Block Models) is a topic model based on Stochastic Block Models (SBM) that models the relationships between nodes (documents) according to groups (topics). TopSBM utilizes SBM to divide a set of documents into topics (groups) and performs topic modeling by estimating the distribution of words within each topic. TopSBM is differentiated from other topic modeling techniques by possessing the ability to model the relationships between topics, a characteristic of SBM; this allows for the identification of interactions and influences between topics, thereby yielding more sophisticated topic modeling results.

[0030] Latent Dirichlet Allocation (LDA) is one of the most widely used models in topic modeling. It assigns words to specific topics for a given document and assumes that the document consists of a mixture of topics. LDA uses probabilistic graphical modeling techniques to extract topics from a set of documents.

[0031] Latent Semantic Analysis (LSA) is a topic modeling technique that extracts topics from a document set through semantic similarity between words. It uses Singular Value Decomposition to compress the document-word matrix into a lower dimension, thereby identifying the relationships between documents and words.

[0032] Non-Negative Matrix Factorization (NMF) is a method for extracting topics by decomposing a document-word matrix into two lower-dimensional matrices. While similar to LSA, NMF has the characteristic of generating matrices without negative values, allowing it to model words as a linear combination of documents and topics.

[0033] The Hierarchical Dirichlet Process (HDP) is an extension of LDA that infers the number of topics from data without the need to specify the number of topics in advance. HDP generates a topic hierarchy using unsupervised learning and can flexibly handle which topic a given document belongs to.

[0034] Unlike LDA, the Biterm Topic Model (BTM) performs topic modeling based on pairs of two words called biterms extracted from documents, rather than the documents themselves. BTM is known to be a useful model for application to documents such as short texts or tweets.

[0035] The keyword extraction unit (100) extracts patent topics, trademark topics, IPC keywords, and industry keywords.

[0036] Multiple patent topics can be extracted from patent disclosure information. Patent topics may be extracted by selecting one of 'bibliographic information', 'title of invention', 'abstract', 'claims', and 'personal information' from the patent disclosure information, or by selecting at least two of these. New patent information may be added to the patent topics, and previously registered patent information may be included. Patent topics can be edited by patent field.

[0037] Multiple trademark topics can be extracted from publicly available trademark information. Trademark topics may be extracted by selecting one of 'bibliographic information', 'trademark name', 'personal name information', or 'designated goods' from the publicly available trademark information, or by selecting at least two of these. New trademark information may be added to trademark topics, and previously registered trademark information may be included. Trademark topics can be edited by trademark field.

[0038] Multiple IPC keywords can be extracted from IPC information. IPC keywords can be extracted by selecting 'IPC code' and / or 'Korean title' from the IPC information. New IPC information may be added to the IPC keywords, and previously registered IPC information may be included. IPC keywords can be edited by technology field.

[0039] Multiple industry keywords can be extracted from KSIC information. Industry keywords can be extracted by selecting any one of the KSIC 'Classification Code', 'Classification Name', or 'Description', or by selecting at least two of these. New industry information may be added to the industry keywords, and previously registered business information may be included. Industry keywords can be edited by industry sector.

[0040] The similarity measurement unit (200) extracts the similarity of given patent topics, trademark topics, IPC keywords, and industry keywords from the keyword extraction unit (100). The similarity measurement unit (200) calculates the difference between probability distributions based on Kullback-Leibler divergence (KL-Divergence). The similarity measurement unit (200) extracts the similarity between technology and product, fused technology information, and new item information.

[0041] The similarity between technology and products is calculated as the difference between the probability distributions of patent topics and trademark topics, and is quantified as similarity. The more similar the probability distributions are, the higher the similarity between patents and trademarks; conversely, the more different they are, the lower the similarity. The similarity value ranges from 0 to 1, and a value closer to 1 indicates that the two distributions are similar.

[0042] Convergence technology information is calculated as the difference between the probability distributions of patent topics and IPC keywords, which is quantified as similarity. The more similar the probability distributions are, the higher the similarity between the patent and the technology; conversely, the more different they are, the lower the similarity. The similarity value ranges from 0 to 1, and a value closer to 1 indicates that the two distributions are similar.

[0043] New item information is calculated as the difference between the probability distributions of trademark topics and industry keywords, which is quantified as similarity. The more similar the probability distributions are, the higher the similarity between patents and technologies; conversely, the more different they are, the lower the similarity. The similarity value ranges from 0 to 1, and a value closer to 1 indicates that the two distributions are similar.

[0044] The display unit (300) displays similarity information calculated from the similarity measurement unit (200) as a network composed of nodes and edges, which can be represented as a bipartite graph.

[0045] The display unit (300) displays the similarity between technology and product as a network composed of patent topic nodes, trademark topic nodes, and calculated similarity edges between patent topics and trademark topics.

[0046] Additionally, the display unit (300) displays fusion technology information as a network composed of patent topic nodes, IPC keyword nodes, and similarity edges calculated between patent topics and IPC keywords.

[0047] Additionally, the display unit (300) displays new item information in a network composed of trademark topic nodes, industry keyword nodes, and edges representing the similarity calculated between trademark topics and industry keywords.

[0048] FIG. 2 is a diagram showing the process of a device (10) providing information on fusion technology and new item information according to one embodiment of the present invention.

[0049] Referring to FIG. 2, the device (10) for providing information on convergence technology and new items can display information on convergence technology and new items by extracting keywords from patent disclosure information, trademark disclosure information, IPC information, and KSIC information and measuring similarity.

[0050] The process of providing information by the fusion technology and new item information providing device (10) of the present invention is described in detail below.

[0051] First, in the keyword extraction step (S10), the keyword extraction unit (100) can extract patent topics and trademark topics from patent disclosure information and trademark disclosure information.

[0052] The keyword extraction unit (100) calculates the word distribution and document distribution based on the TopSBM model to determine the number of topics for patent topics and / or trademark topics. The topics thus determined consist of keywords representing each topic and probability values ​​calculated from the frequency of the keywords. That is, the corresponding topics can be interpreted through the representative keywords.

[0053] According to an embodiment, when a user enters the patent application date (AD) in the patent search DB 'KIPIRS' as the period '2020.01.01~2022.12.31' and enters 'APPLE*Inc.' as the patent applicant (AP) to search for US patent information in the 'Title of Invention' and 'Abstract' categories, 1,790 patent disclosure information results can be obtained. The keyword extraction unit (100) can extract patent keywords from the 1,790 patent disclosure information, analyze the patterns of the extracted patent keywords, and group patent keywords with similar meanings to extract 7 patent topics.

[0054] FIG. 3 is a diagram showing seven patent topics extracted from the keyword extraction unit (100) of FIG. 2 and patent keywords included in each patent topic.

[0055] Referring to Figure 3, it can be seen that 'data' is a word representing patent topic 4 and is the keyword that appears with the highest frequency in the patent information.

[0056] Then, when a user enters 'APPLE*Inc.' in the trademark applicant (AP) field in the trademark search DB 'KIPIRS' and searches for US trademark disclosure information, 140 trademark rights information results can be obtained. The keyword extraction unit (100) can extract trademark keywords from the 140 trademark rights information, analyze the patterns of the extracted keywords, and group keywords with similar meanings to extract two trademark topics.

[0057] FIG. 4 is a diagram showing 23 trademark topics extracted from the keyword extraction unit (100) of FIG. 2 and trademark keywords included in each trademark topic.

[0058] Referring to Figure 4, it can be seen that 'electronic' is a word representing trademark topic 2 and is a keyword that appears with high frequency in trademark information.

[0059] IPC keywords and industry keywords can be extracted from the keyword extraction unit (100) by the above-described method and / or process.

[0060] Figure 5 shows the probability that the title of the invention 'weather user interface' of the 531st patent disclosure (US2023-0078135 A1) among the 1,790 patent disclosures of Figure 2 is included in each patent topic.

[0061] Referring to Figure 5, the 531st patent disclosure information 'weather user interface' is most likely to be included in patent topic 3 (topic3) among the 7 patent topics.

[0062] The keyword extraction unit (100) can estimate the probability value to be included in each keyword using Gibbs sampling, which is one of the Markov Chain Monte Crario (MCMC) methods.

[0063] Below, the estimation process using Gibbs sampling is explained using the 531st patent disclosure information, 'weather user interface', as an example. First, all keywords constituting the patent keywords extracted from the 531st patent disclosure information are assigned to arbitrary patent topics. Then, a model is trained on all possible patent topics, ranging from the case where all patent keywords are assigned to a single patent topic to the case where each patent keyword is assigned to an individual patent topic, to calculate the probability distribution of each topic. This process is repeated, and all patent keywords are adjusted to be assigned to appropriate patent topics.

[0064] Figure 6 is a diagram showing the probability that the first trademark keyword 'IPHONE 11 (Trademark name)' among the 140 trademark information items of Figure 2 is included in each trademark topic by the method described above.

[0065] Referring to Figure 6, the first trademark keyword 'IPHONE 11 (Trademark name)' has the highest probability of being included in trademark topic 1 among 23 trademark topics.

[0066] Patent topics, trademark topics, IPC keywords, and industry keywords extracted from the keyword extraction unit (100) can subsequently be extracted as similarity between technology and product, fusion technology information, and new item information through the similarity measurement unit (200) in the similarity measurement step (S20).

[0067] The similarity measurement unit (200) calculates the difference between the probability distributions between the patent topic and the trademark topic to extract the similarity between the technology and the product.

[0068] Below, the method by which the similarity measuring unit (200) calculates similarity is explained.

[0069] FIG. 7 is a diagram illustrating a method for calculating similarity between technology and product according to an embodiment of the present invention, and FIG. 8 is a diagram showing the result value calculated by the method of FIG. 7. Hereinafter, for convenience of explanation, patent keywords and trademark keywords extracted from the keyword extraction unit (100) are referred to as ‘appearing words’.

[0070] The similarity measurement unit (200) converts the frequency of words appearing in the patent topic and trademark topic, respectively, into a vector, and can extract the similarity between the technology and the product using the angle between the two vectors.

[0071] First, referring to FIG. 7, the similarity measurement unit (200) can extract all words appearing in the nth patent topic and the nth trademark topic among the patent topics and trademark topics extracted from the keyword extraction unit (100), and represent them as a word-frequency matrix. According to the embodiment, the rows in the matrix can be represented as patent topics and trademark topics, and the columns can be represented as 'appearing words'. The similarity measurement method is explained using the previously described 'weather user interface' and 'IPHONE 11' as examples. The words appearing in the patent topic containing 'weather user interface' and the trademark topic containing 'IPHONE 11' are "patent, invention, protection, trademark, registration, product, brand, logo, design", and five words "patent, invention, protection, trademark, registration" appeared in the patent topic, and five words "trademark, product, brand, logo, design" appeared in the trademark topic. The similarity measurement unit (200) converts the word-frequency matrices appearing in the patent topic containing 'weather user interface' and the trademark topic containing 'IPHONE 11' into vectors, respectively.

[0072] And with reference to FIG. 8, the similarity measuring unit (200) can calculate cosine similarity using the angle between two vectors. Cosine similarity is calculated using the dot product of two vectors and the magnitude of each vector, and the dot product of two vectors is divided by the product of the magnitudes of the two vectors. The dot product is calculated by multiplying each component of the two vectors and then adding them, and the magnitude of the vector is calculated by adding the squares of each component of the vector and then taking the square root. Cosine similarity has a range from 1 to 1, and if the two vectors point in exactly the same direction, the cosine value is 1, and if they point in exactly opposite directions, the cosine value is -1.

[0073] By the method described above, the inner product of the patent topic vector containing 'weather user interface' and the trademark topic vector containing 'IPHONE 11' is calculated as 1, the magnitude of the patent topic vector is calculated as 2.236, and the magnitude of the trademark topic vector is calculated as 2.236. The similarity between the patent topic vector and the trademark topic vector is calculated as 0.447. Thus, the similarity between the technology and product of 'weather user interface' and 'IPHONE 11' is extracted as 0.4425.

[0074] FIG. 9 is a drawing for explaining a method for calculating similarity between technology and product according to another embodiment of the present invention, and FIG. 10 is a drawing showing a result value calculated by the method of FIG. 9.

[0075] Referring to FIGS. 9 and 10, the similarity measuring unit (200) can extract weighted similarity between technology and product by reflecting probability values ​​to each of the patent topic containing 'weather user interface' of FIG. 7 and the trademark topic containing 'IPHONE 11'.

[0076] First, referring to Fig. 9, the patent topic and trademark topic of the word-frequency matrix of Fig. 7 can each be represented by a probability value. The probability value is calculated based on the frequency with which words appearing in each of the n patent topic and n trademark topic appear in the entire document. The sum of the probability values ​​for each of the n patent topic and n trademark topic is 1.

[0077] Referring to FIG. 10, the similarity measurement unit can calculate cosine similarity using the inner product of the two vectors calculated in FIG. 8 and the magnitude of the two topic vectors reflecting the probability value in FIG. 9. Cosine similarity is calculated by dividing the inner product of the two vectors by the product of the magnitudes of the two vectors. The magnitude of the vector is calculated by multiplying the square of each component of the vector by the probability value of each component, adding them, and taking the square root. Cosine similarity has a range from 1 to 1, and if the two vectors point in exactly the same direction, the cosine value is 1, and if they point in exactly opposite directions, the cosine value is -1.

[0078] By the method described above, the magnitude of the patent topic vector with probability values ​​reflected is calculated as 1, and the magnitude of the trademark topic vector with probability values ​​reflected is calculated as 1. Then, the weighted similarity between the patent topic vector and the trademark topic vector is calculated as 1. Thus, the weighted similarity between the technology and product of 'weather user interface' and 'IPHONE 11' is extracted as '1'.

[0079] FIG. 11 is a drawing showing the similarity between the technology and product of FIG. 8, FIG. 10 is a drawing showing fusion technology information displayed in the display step (S30) according to one embodiment of the present invention, and FIG. 11 is a drawing showing new item information displayed in the display step (S30) according to one embodiment of the present invention.

[0080] Referring to Fig. 9, the similarity measurement results of the 'weather user interface' and 'IPHONE 11' in Fig. 8 are displayed as a network consisting of a patent topic node containing the weather user interface, a trademark topic node containing 'IPHONE 11', and their similarity edges, allowing for the visualization of similarity between the technology and the product.

[0081] Referring to Fig. 10, the results of measuring the similarity between a patent topic node containing 'weather user interface' and an IPC keyword of the 'weather user interface' patent are displayed as a network composed of a patent topic node containing 'weather user interface', IPC keyword (G06F3 / 0362, G06F3 / 0487, G06F3 / 0484) nodes and their similarity edges, and the convergence technology information can be visualized.

[0082] Referring to Fig. 11, the results of measuring the similarity between a trademark topic containing 'IPHONE 11' and an industry keyword are displayed as a network consisting of a trademark topic node containing 'IPHONE 11', an industry keyword node, and their similarity edges, allowing for the visualization of new item information.

[0084] Although the present invention has been described in detail using preferred embodiments, the scope of the invention is not limited to specific embodiments and should be interpreted by the appended claims. Furthermore, those skilled in the art will understand that many modifications and variations are possible without departing from the scope of the invention. Explanation of the symbols

[0085] 10: Fusion Technology and New Item Information Provision Device 100: Keyword Extraction Section 200: Similarity Measurement Unit 300: Display section S10: Keyword extraction step S20: Similarity measurement step S30: Display stage

Claims

Claim 1 A keyword extraction unit that extracts multiple patent topics from patent disclosure information, extracts multiple trademark topics from trademark disclosure information, extracts multiple IPC keywords from IPC information, and extracts multiple industry keywords from KSIC information; a similarity measurement unit that calculates the difference between probability distributions based on Kullback-Leibler Divergence (KL-Divergence) between multiple technology keywords included in each of the patent topics and multiple product keywords included in each of the trademark topics, the similarity of the patent topics, the similarity of the IPC keywords, and the similarity of the industry keywords, and extracts similarity between technology and product, convergence technology information, and new item information; A device for providing information on fused technology and new items, comprising a display unit that displays the similarity between the technology and the product as a network composed of similarity edges between a patent topic node and a trademark topic node, and the fused technology information as a network composed of similarity edges between a patent topic node and an IPC keyword node, wherein the fused technology information is calculated as the difference between the probability distributions of the patent topic and the IPC keyword, and the new item information is calculated as the difference between the probability distributions of the trademark topic and the industry keyword. Claim 2 In claim 1, the keyword extraction unit is a convergence technology and novel item information providing device capable of extracting the patent topic and the trademark topic using any one of TopSBM (Topic Models based on Stochatic Block Models), Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), Non-Negative Matrix Factorization (NMF), Hierarchical Dirichlet Process (HDP), and Biterm Topic Model (BTM). Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 A method for providing information on convergent technology and new items, comprising: a keyword extraction step through a keyword extraction unit, which extracts multiple patent topics from patent disclosure information, extracts multiple trademark topics from trademark disclosure information, extracts multiple IPC keywords from IPC information, and extracts multiple industry keywords from KSIC information; a similarity measurement step through a similarity measurement unit, which calculates the difference between probability distributions based on Kullback-Leibler divergence (KL-Divergence) between multiple technology keywords included in each of the patent topics and multiple product keywords included in each of the trademark topics, the similarity of the patent topics, the similarity of the IPC keywords, and the similarity of the industry keywords, in order to extract similarity between technology and products, convergent technology information, and new item information; and a display step through a display unit, which displays the similarity between technology and products as a network composed of similarity edges between a patent topic node and a trademark topic node, and the convergent technology information as a network composed of similarity edges between a patent topic node and an IPC keyword node. Claim 7 In claim 6, the keyword extraction step extracts the patent topic and the trademark topic using any one of TopSBM (Topic Models based on Stochatic Block Models), Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), Non-Negative Matrix Factorization (NMF), Hierarchical Dirichlet Process (HDP), and Biterm Topic Model (BTM), a fusion technology and a method for providing new item information. Claim 8 delete Claim 9 delete Claim 10 delete

Images