KR102300352B1 - Method, apparatus and system for determining similarity of patent documents based on importance score - Google Patents

Abstract

A method, apparatus and system for determining the degree of similarity of patent documents are provided. The control method of the system for determining the degree of similarity of the patent document includes: acquiring, by a server, a target patent document; obtaining, by the server, at least one word based on the target patent document; obtaining, by the server, an importance score of the obtained at least one word; obtaining, by the server, the similar patent document by inputting the target patent document into a first artificial intelligence model; obtaining, by the server, a plurality of sentences included in the target patent document, and acquiring a plurality of sentences included in the similar patent document; obtaining, by the server, a first sentence among a plurality of sentences included in the target patent document, and acquiring a second sentence among a plurality of sentences included in the similar patent document; inputting, by the server, the first sentence and the second sentence into a second artificial intelligence model, and obtaining evaluation results for the first sentence and the second sentence; and generating, by the server, a prior art search report for the target patent document based on the evaluation result.

Description

Method, apparatus and system for judging similarity of patent documents based on importance score

The present invention relates to a method, apparatus and system for determining the similarity of patent documents.

With the development of the 4th industry, the value of intellectual property rights is increasing. Accordingly, many people are trying to protect their technology and acquire the right to the technology, and interest in patent application for technology is increasing.

Meanwhile, in order to apply for a patent, prior art research is performed to determine whether one's own technology can be patented, and prior art research can be performed by searching various patent documents published in the past.

The most important thing in conducting a prior art search is to determine whether a similar patent document that can deny the inventive step of the target patent document exists.

However, the amount of patent documents is huge and it is virtually impossible to analyze all patent documents published in the past due to time constraints. It is a reality to obtain similar patent documents.

However, obtaining a similar patent document through a search-type input, etc. is often dependent on the ability of a person who conducts a prior art search, and thus, in many cases, a stable result of a prior art search is not guaranteed.

Therefore, the need for a prior art research method that can guarantee stable results with a small investment of time is emerging.

Laid-open Patent Publication No. 10-2018-0110713, 2018.10.11

The problem to be solved by the present invention is to provide a method, apparatus and system for determining the degree of similarity of patent documents.

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

A control method of a system for determining a degree of similarity of a patent document according to an aspect of the present invention for solving the above-described problem includes, by a server, acquiring a target patent document; obtaining, by the server, at least one word based on the target patent document; obtaining, by the server, an importance score of the obtained at least one word; obtaining, by the server, the similar patent document by inputting the target patent document into a first artificial intelligence model; obtaining, by the server, a plurality of sentences included in the target patent document, and acquiring a plurality of sentences included in the similar patent document; obtaining, by the server, a first sentence among a plurality of sentences included in the target patent document, and acquiring a second sentence among a plurality of sentences included in the similar patent document; inputting, by the server, the first sentence and the second sentence into a second artificial intelligence model, and obtaining evaluation results for the first sentence and the second sentence; and generating, by the server, a prior art search report for the target patent document based on the evaluation result.

In this case, the acquiring of the word may include: acquiring, by the server, a plurality of words by analyzing morphemes based on the target patent document; determining, by the server, an erroneous word among the plurality of acquired words; correcting, by the server, the error word; and obtaining, by the server, at least one set of compound noun phrases based on the degree of association between the plurality of words.

In this case, the step of obtaining the importance score may include: obtaining, by the server, a word to be calculated for the importance score from the target patent document; The server includes the first detailed importance of the word in the entire patent document, the second detailed importance of the word in the patent classification information corresponding to the technical field information of the target patent document, and the word in the entire patent document calculating at least one detailed importance among third detailed importance levels of the searched patent document; and calculating, by the server, the importance score of the word based on at least one of the first detailed importance level, the second detailed importance level, and the third detailed importance level.

In this case, the control method may include: acquiring, by the server, the entire patent document that is the basis of the thesaurus; analyzing, by the server, morphemes for each of the entire patent documents; converting a word included in each of the entire patent documents into a word vector based on the result of the morpheme analysis; calculating, by the server, a degree of similarity between the word vectors; and grouping, by the server, the word corresponding to the word vector into a similar word group based on the degree of similarity.

In this case, the determining of the erroneous word may include: matching semantic information for each of the obtained plurality of words; obtaining an accuracy score for each of the plurality of words based on the matched semantic information and the target patent document; and determining a word having the accuracy score equal to or less than a preset value as an erroneous word. obtaining a plurality of weights for each of the obtained pieces of semantic information; obtaining a weight having the highest degree of relevance with the target patent document from among the plurality of weights; and matching semantic information corresponding to a weight having the highest degree of relevance with the target patent document to the erroneous word, wherein the acquiring of the at least one complex noun phrase set includes: obtaining a plurality of words for a sentence; obtaining a plurality of compound noun phrase set candidates obtained by combining the plurality of words; obtaining a frequency in which a compound noun phrase set identical to the obtained plurality of compound noun phrase set candidates is included in the target patent document; and determining a compound noun phrase set candidate whose frequency is equal to or greater than a preset frequency as a compound noun phrase set. and, the compound noun phrase set candidate includes order information and spacing information of words included in the compound noun phrase set candidate.

In this case, the step of calculating the detailed importance of the word includes: a first appearance ratio of the number of appearances of the word in the entire patent document compared to the total number of words in the entire patent document, and the total number of sentences in the entire patent document calculating the first detailed importance level based on a second appearance ratio of the number of appearance sentences in which the word appears in the sentences of the patent document; A third appearance ratio of the number of appearances of the word in the patent classification information compared to the total number of words in the patent classification information and the number of appearance sentences in which the word appears in the entire patent document compared to the total number of sentences in the entire patent document calculating the second detailed importance level based on a fourth appearance ratio of ; and calculating an influence value of each of the searched patent documents based on the reference information of each of the searched patent documents, and calculating a third detailed importance level of the searched patent document using the influence value, wherein the first The calculating of the detailed importance may include calculating the first detailed importance using Equation 1 below, and the calculating of the second detailed importance may include calculating the second detailed importance using the following Equation. including the steps of

<Equation 1>

<Equation 2>

Here, W1 is the first detailed importance level, the wpw is the number of appearances of the word in the entire patent document, the WPW is the total number of words in the entire patent document, and the wps is the number of occurrences of the word in the entire patent document. is the number of sentences in which a word appears, the WPS is the total number of sentences in the entire patent document, a1 is an adjustment constant of the second appearance ratio, W2 is a second detailed importance, and ipcw is the patent classification is the number of occurrences of the word in the information, the IPCW is the total number of words in the patent classification information, the ipcs is the number of occurrences of the word among the sentences of the entire patent document, and the IPCS is the total number of words in the entire patent document. It is the total number of sentences, and a2 is an adjustment constant of the fourth appearance rate.

In this case, the acquiring of the similar patent document may include: acquiring a plurality of words included in the target patent document; obtaining a plurality of target patent clusters by clustering the obtained plurality of words; A plurality of midpoints for each of the plurality of target patent clusters are obtained, and the location of the target patent document is determined based on the plurality of midpoints for each of the plurality of target patent clusters and the number of words included in the target patent cluster. to do; obtaining a plurality of words included in the patent document, and clustering the obtained plurality of words to obtain a plurality of patent clusters; obtaining a plurality of midpoints for each of the plurality of patent clusters, and determining the position of the patent document based on the plurality of midpoints for each of the plurality of patent clusters and the number of words included in the patent cluster; and when the position of the target patent document and the position of the patent document are within a preset distance, determining the patent document as the similar patent document, wherein the evaluation result of the first sentence and the second sentence The acquiring may include: acquiring a similarity score and a dissimilarity score of the first sentence and the second sentence, respectively; determining that the first sentence and the second sentence are unrelated sentences when the dissimilarity score is equal to or greater than a preset score; acquiring at least one word not included in the second sentence among words included in the first sentence when the similarity score is equal to or greater than a preset score; obtaining at least an importance score for each of at least one word not included in the second sentence, and determining whether a word with a predetermined importance score or higher among the at least one obtained importance score exists; determining the first sentence and the second sentence as matching sentences when there is no word equal to or greater than a preset importance score; and determining that the first sentence and the second sentence are mismatched sentences when there is a word equal to or greater than the preset importance score.

In this case, the step of converting the word vector includes converting the word into a word vector through Word2Vec learning based on the patent document, and calculating the similarity between any two word vectors among the word vectors. calculating the distance and calculating the calculated distance as a degree of similarity, wherein the grouping into the similar word group includes determining whether the degree of similarity between any two word vectors among the word vectors is less than a preset reference degree of similarity, and if the similarity between any two word vectors among the word vectors is less than a preset reference similarity, grouping two words corresponding to the two word vectors into the similar word group, and obtaining the entire patent document includes: and checking whether any one of the acquired patent documents satisfies the noise document condition, and removing the patent document satisfying the noise document condition.

Other specific details of the invention are included in the detailed description and drawings.

According to the above-described embodiment of the present invention, a prior art research method can be provided in which a user can be guaranteed a stable result with a small investment of time.

Effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a system diagram according to an embodiment of the present invention.
2 is a flowchart illustrating a method of obtaining a prior art research report according to an embodiment of the present invention.
3 is a flowchart illustrating a method of acquiring a word according to an embodiment of the present invention.
4 is a flowchart illustrating a method of calculating an importance score according to an embodiment of the present invention.
5 is a flowchart illustrating a method of generating a thesaurus according to an embodiment of the present invention.
6A to 6C are flowcharts specifically illustrating a method for obtaining a word according to an embodiment of the present invention.
7 is a flowchart specifically illustrating a method of calculating an importance score according to an embodiment of the present invention.
8A and 8B are flowcharts illustrating a method for obtaining a similar patent document and a method for determining a similar sentence according to an embodiment of the present invention.
9 is a flowchart illustrating a specific method of generating a thesaurus according to an embodiment of the present invention.
10 is an exemplary diagram for explaining a process of deriving an evaluation result using a first artificial intelligence model and a second artificial intelligence model according to an embodiment of the present invention.
11 is a block diagram of an apparatus according to an embodiment of the present invention.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and those of ordinary skill in the art to which the present invention pertains. It is provided to fully understand the scope of the present invention to those skilled in the art, and the present invention is only defined by the scope of the claims.

The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present invention. As used herein, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other components in addition to the stated components. Like reference numerals refer to like elements throughout, and "and/or" includes each and every combination of one or more of the recited elements. Although "first", "second", etc. are used to describe various elements, these elements are not limited by these terms, of course. These terms are only used to distinguish one component from another. Accordingly, it goes without saying that the first component mentioned below may be the second component within the spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used herein will have the meaning commonly understood by those of ordinary skill in the art to which this invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless specifically defined explicitly.

As used herein, the term “unit” or “module” refers to a hardware component such as software, FPGA, or ASIC, and “unit” or “module” performs certain roles. However, “part” or “module” is not meant to be limited to software or hardware. A “unit” or “module” may be configured to reside on an addressable storage medium or to reproduce one or more processors. Thus, by way of example, “part” or “module” refers to components such as software components, object-oriented software components, class components and task components, processes, functions, properties, Includes procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. Components and functionality provided within “parts” or “modules” may be combined into a smaller number of components and “parts” or “modules” or as additional components and “parts” or “modules”. can be further separated.

Spatially relative terms "below", "beneath", "lower", "above", "upper", etc. It can be used to easily describe the correlation between a component and other components. A spatially relative term should be understood as a term that includes different directions of components during use or operation in addition to the directions shown in the drawings. For example, when a component shown in the drawing is turned over, a component described as “beneath” or “beneath” of another component may be placed “above” of the other component. can Accordingly, the exemplary term “below” may include both directions below and above. Components may also be oriented in other orientations, and thus spatially relative terms may be interpreted according to orientation.

In this specification, a computer means all types of hardware devices including at least one processor, and may be understood as encompassing software configurations operating in the corresponding hardware device according to embodiments. For example, a computer may be understood to include, but is not limited to, smart phones, tablet PCs, desktops, notebooks, and user clients and applications running on each device.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Each step described in this specification is described as being performed by a computer, but the subject of each step is not limited thereto, and at least a portion of each step may be performed in different devices according to embodiments.

1 is a system diagram according to an embodiment of the present invention.

The patent document similarity determination system according to the present invention includes a server 10 and an electronic device 20 .

The server 10 is configured to obtain a target patent document, obtain a similar patent document from the acquired target patent document, and determine the similarity between a sentence of the target patent document and a sentence of the similar patent document.

As an embodiment, the server 10 may receive a target patent document from the electronic device 20 or acquire a target patent document from an external server.

In another embodiment, the server 10 may obtain a plurality of words from a target patent document or a similar patent document, and obtain an importance score of the obtained word. In addition, the server 10 may obtain a thesaurus from all patent documents and use it to determine the importance score of a word.

In another embodiment, the server 10 may obtain a similar patent document for the target patent document, obtain a similar sentence, and generate a prior art search report.

In the present specification, a patent document is a concept including a target patent document and a similar patent document, and may be a document about technical contents submitted by an applicant to obtain a patent registration at each country's patent office. However, the present invention is not limited thereto, and the patent document may be understood as a concept including various documents including technical contents, such as a job invention for a patent application, and a thesis. According to an embodiment, the target patent document may be at least one of a job invention book and a thesis for a patent application, and the similar patent document may be at least one of a job invention book, a thesis, and a patent application for a patent application.

The electronic device 20 is configured to provide a patent document to the server 10 . The electronic device 200 according to the present invention may be implemented as a smartphone, but this is only an embodiment, and a smartphone, a tablet personal computer (PC), a mobile phone, a video phone, e-book reader, desktop PC, laptop PC, netbook computer, workstation, server, PDA (personal digital assistant), PMP (portable multimedia) player) or a wearable device.

2 is a flowchart illustrating a method of obtaining a prior art research report according to an embodiment of the present invention.

In step S110, the server 10 may obtain a target patent document.

In step S120, the server 10 may acquire at least one word based on the target patent document.

In step S130, the server 10 may obtain an importance score of the obtained at least one word.

In step S140, the server 10 may obtain a similar patent document by inputting the target patent document into the first artificial intelligence model.

Specifically, the server 10 may obtain a plurality of words from a target patent document or a similar patent document, and obtain an importance score of the obtained word. In addition, the server 10 may obtain a thesaurus from all patent documents and use it to determine the importance score of a word.

In one embodiment, the server 10 analyzes each of the plurality of patent documents based on the morpheme analysis of each of the plurality of patent documents stored in the first database, and based on the morpheme analysis of the target patent document, the target patent After analyzing the document, it is possible to obtain a similar patent document with a high degree of relevance to the target patent document.

In step S150, the server 10 may obtain a plurality of sentences included in the target patent document, and may acquire a plurality of sentences included in the similar patent document.

In step S160, the server 10 may acquire a first sentence among a plurality of sentences included in the target patent document, and acquire a second sentence among a plurality of sentences included in the similar patent document.

In step S170 , the server 10 may input the first sentence and the second sentence into the second artificial intelligence model to obtain evaluation results for the first sentence and the second sentence.

In step S180, the server 10 may generate a prior art search report for the target patent document based on the evaluation result. Specifically, the prior art search report for the target patent document may be generated based on the evaluation results of the first sentence and the second sentence. In one embodiment, when the evaluation result for the first sentence and the second sentence is obtained in the form of a score, the server 10 determines the first sentence in which the evaluation result score for the first sentence and the second sentence is equal to or greater than a preset score. and a prior art research report may be generated based on the second sentence corresponding to the first sentence.

On the other hand, the first artificial intelligence model is an artificial intelligence model based on a convolutional deep neural network (CNN) learned by inputting a plurality of patent documents as learning data, and the second artificial intelligence model is a plurality of patent documents and the It may be an artificial intelligence model based on two patent documents among a plurality of patent documents and a Bi-LSTM model learned based on prior art research reports on the two patent documents. However, the present invention is not limited thereto, and it goes without saying that various artificial intelligence models may be applied to the present invention. For example, a model such as a Deep Neural Network (DNN), a Recurrent Neural Network (RNN), or a Bidirectional Recurrent Deep Neural Network (BRDNN) may be used as the AI model, but is not limited thereto.

In this case, convolutional deep neural networks (CNNs) are a type of multilayer perceptrons designed to use minimal preprocessing. A convolutional neural network consists of one or several convolutional layers and general artificial neural network layers on top of it, and additionally utilizes weights and pooling layers. Thanks to this structure, the convolutional neural network can fully utilize the input data of the two-dimensional structure. In addition, convolutional neural networks can be trained via standard backpropagation. Convolutional neural networks are easier to train than other feedforward neural network techniques and have the advantage of using fewer parameters. A convolutional neural network extracts features from an input image by alternately performing convolution and subsampling on the input image.

A convolutional neural network includes several convolution layers, several subsampling layers (Subsampling layer, Lacal pooling layer, Max-Pooling layer), and a fully connected layer. The convolution layer is a layer that performs convolution on an input image. In addition, the subsampling layer is a layer that extracts the maximum value locally from the input image and maps it to a 2D image, and subsampling is performed by enlarging the local area.

In the convolution layer, information such as the kernel size, the number of kernels to be used (ie, the number of maps to be generated), and a weight table to be applied during the convolution operation are required. For example, suppose that the size of the input image is 32x32, the size of the kernel is 5x5, and the number of kernels to be used is 20. In this case, if a 5×5 kernel is applied to a 32×32 input image, it is impossible to apply the kernel to two pixels above, below, left, and right of the input image. If the kernel is placed on the input image and then convolution is performed, the result value of '-8' is determined as the value of the pixel corresponding to the center element of the kernel among the pixels of the input image included in the kernel. am. Therefore, when convolution is performed by applying a 5×5 kernel to an input image having a size of 32×32, a map of size 28×28 is generated. Previously, since it was assumed that the total number of kernels to be used is 20, in the first convolutional layer, a total of 20 maps with a size of 28×28 are generated.

In the subsampling layer, information on the size of a kernel to be subsampled and information on whether to select a maximum value or a minimum value among values in the kernel region are required.

In addition, deep neural networks (DNNs) are artificial neural networks (ANNs) composed of a plurality of hidden layers between an input layer and an output layer.

In this case, the structure of the deep neural network may be composed of a perceptron. A perceptron consists of several inputs, one processor, and one output value. The processor multiplies each input value by a weight, and then sums the input values multiplied by the weight. Then, the processor outputs one output value by substituting the summed value into the activation function. If a specific value is desired as the output value of the activation function, the weight multiplied by each input value may be corrected, and the output value may be recalculated using the modified weight. In this case, each perceptron may use a different activation function. In addition, each perceptron receives the outputs from the previous layer as input, and then uses the activation function to obtain the output. The obtained output is transferred to the input of the next layer. Through the process as described above, some output values can be finally obtained.

In addition, returning to the description of the deep learning technique, a Reccurent Neural Network (RNN) refers to a neural network in which connections between units constituting an artificial neural network constitute a directed cycle. Unlike forward neural networks, recurrent neural networks can utilize the memory inside the neural network to process arbitrary inputs.

Deep Belief Networks (DBN) is a generative graphical model used in machine learning, and in deep learning, it means a deep neural network composed of multiple layers of latent variables. There is a connection between layers, but there is no connection between units within a layer.

The deep trust neural network can be used for pre-learning due to the nature of the generative model, and after learning the initial weights through pre-learning, the weights can be fine-tuned through backpropagation or other discrimination algorithms. This feature is very useful when the training data is small, because the smaller the training data, the stronger the effect of the initial value of the weight on the resulting model. The pre-learned initial weight value is closer to the optimal weight compared to the arbitrarily set initial weight value, which enables performance and speed improvement in the fine-tuning stage.

The above-described artificial intelligence and its learning method have been described for illustrative purposes, and the artificial intelligence and its learning method used in the embodiments described below are not limited. For example, all kinds of artificial intelligence technology and a learning method thereof that a person skilled in the art can apply to solve the same problem may be utilized to implement the system according to the disclosed embodiment.

3 is a flowchart illustrating a method of acquiring a word according to an embodiment of the present invention.

In step S210, the server 10 may obtain a plurality of words by analyzing morphemes based on the target patent document.

In one embodiment, the server 10 may perform morpheme analysis of the target patent document by using the Mecab morpheme analyzer. However, the present invention is not limited thereto, and in some cases, various morpheme analyzers such as Okt, Komoran, Hannanum, and Kkma morpheme analyzers may be used. Furthermore, it goes without saying that the server 10 may use various morpheme analyzers according to the language of the target patent document to be analyzed. Meanwhile, in the present invention, a method of analyzing a morpheme of a target patent document has been described, but it goes without saying that the server 10 can perform morpheme analysis on various patent documents as well as the target patent document.

In step S220 , the server 10 may determine an erroneous word among the plurality of acquired words.

In an embodiment, when the plurality of acquired words do not satisfy a preset condition, the server 10 may determine the corresponding word as an erroneous word.

In step S230, the server 10 may correct the error word.

In step S240 , the server 10 may acquire at least one set of compound noun phrases based on the degree of association between a plurality of words.

4 is a flowchart illustrating a method of calculating an importance score according to an embodiment of the present invention.

In step S310 , the server 10 may obtain a word that is a target for calculating an importance score from the target patent document.

In step S320, the server 10 determines that the first detailed importance of the word in the entire patent document, the second detailed importance of the word in the patent classification information corresponding to the technical field information of the target patent document, and the word in the entire patent document are At least one detailed importance among the third detailed importance of the included search patent document may be calculated.

In operation S330 , the server 10 may calculate an importance score of the word based on one or more of the first detailed importance, the second detailed importance, and the third detailed importance.

Specifically, the server 10 may extract only nouns by performing morphological analysis on the target patent document. In this case, it should be noted that the type of morpheme analysis method is not limited as long as the server 10 extracts the noun from the target patent document.

Then, the server 10 may determine the technical field information of the target patent document based on the extracted noun. At this time, the server 10 may compare the word data for each technical field information stored in advance with the extracted nouns, and as a result of the comparison, determine the technical field information including the most extracted nouns as the technical field information of the target patent document.

Thereafter, the server 10 may exclude a word set as a stop word among words included in the target patent document from the word to be calculated for the importance score. Specifically, the server 10 reads stopword data for each technical field information corresponding to the technical field information of the target patent document, and words included in the read stopword data for each technical field information among the extracted words included in the target patent document may be excluded from the word to be calculated for the importance score.

Here, the stop word may mean a word frequently used in the corresponding technical field but not having a technical meaning. For example, although a computer is frequently used in the field of IT technology, it may be defined as an obsolete word because it does not have a technical meaning related to IT technology.

Accordingly, the server 10 may obtain a word that is not a stopword among the words extracted from the target patent document as a word to be calculated for the importance score.

5 is a flowchart illustrating a method of generating a thesaurus according to an embodiment of the present invention.

In step S410, the server 10 may acquire the entire patent document that is the basis of the thesaurus. Specifically, the server 10 may acquire the entire patent document by acquiring a patent document that is published as time elapses in real time. In this case, the server 10 may receive the latest published patent document from the external server or the electronic device 20 through communication and store it in the memory 104 .

In an embodiment, the server 10 may check whether any one patent document among all the acquired patent documents satisfies the noise document condition. Specifically, the server 10 may remove a patent document that satisfies the noise document condition from all patent documents. Here, the noise document condition is to check whether the patent document is a noise document, if there is a space larger than a preset reference size in each of the entire patent documents, if the same character is repeated more than a preset number of times, text in the entire patent document This is a case in which the ratio between and Lee Ji-mi is greater than or equal to the preset reference ratio.

That is, the server 10 may check whether the entire patent document is noise or not. The server 10 may remove a noise document that satisfies a noise document condition.

In step S420, the server 10 may analyze morphemes for each of the entire patent documents.

In step S430, the server 10 may convert a word included in each of the entire patent document into a word vector based on the result of the morpheme analysis.

In this case, the server 10 may convert the plurality of word characteristics of the above-described words into a word vector by projecting the plurality of word characteristics into a multidimensional real space space and vectorizing the same. In an embodiment, the server 10 may convert a word vector to a word using Word2vec learning.

In addition, the server 10 can express a word in a vector space of about 200 to 300 dimensions, and for learning, a word with a continuous bag of words (CBOW) that predicts a target word based on the direction of meaning created by the surrounding words for learning A skip-gram model that predicts words that can come around can be used.

In this case, the distance between the two word vectors may indicate the similarity between words corresponding to each of the two word vectors, and the direction of the word vector may indicate the meaning in the patent document.

In step S440 , the server 10 may calculate a degree of similarity between word vectors.

Specifically, the server 10 may calculate the cosine similarity between two word vectors or calculate the similarity between the two word vectors by normalizing the cosine similarity.

For example, the server 10 may calculate the similarity between the two word vectors by using the cosine value of the angle between the two word vectors in real space. Also, the server 10 may calculate the similarity between the two word vectors by normalizing the cosine similarity value between the two word vectors to have a range between 0 and 1.

That is, the server 10 may convert all words included in the entire patent radio into word vectors, calculate distances between all converted word vectors as cosine similarities, and calculate the calculated distances as similarities.

Meanwhile, according to various embodiments of the present disclosure, the server 10 may perform a post-processing process of a method for generating a thesaurus of a patent document.

Specifically, the server 10 may check whether a word corresponding to the word vector is included in the similar word group based on the degree of similarity. Specifically, the server 10 checks whether the similarity between any two word vectors is less than a preset reference similarity, and if the similarity between any two word vectors is less than the preset reference similarity, the words corresponding to the two word vectors are the same similar words It can be confirmed to be included in the group.

Thereafter, the server 10 may group the word corresponding to the word vector into a similar word group based on the degree of similarity.

Specifically, when it is confirmed that the words respectively corresponding to the two word vectors are included in the same synonym group, the server 10 may include the two words in the same synonym group.

On the other hand, even if the similarity between any two word vectors is less than the preset reference similarity, when the similarity between the word vectors corresponding to the words previously included in the similar word group exceeds the preset maximum similarity, the similarity is set in advance. Words corresponding to two word vectors having less than similarity may not be included in the corresponding similar word group.

Through this, the server 10 can prevent a phenomenon in which the similarity group is expanded and the similarity between any two words is reduced.

6A to 6C are flowcharts specifically illustrating a method for obtaining a word according to an embodiment of the present invention.

Specifically, as shown in FIG. 6A , in step S505 , the server 10 may match semantic information for each of a plurality of acquired words.

In this case, the semantic information may mean an intent for a word. The server 10 may perform natural language processing using an artificial intelligence model to obtain semantic information about a word.

Specifically, the artificial intelligence model includes a natural language understanding unit, and the natural language understanding unit may grasp an entity and the intent of a word included in a sentence based on the sentence analysis result, and further, the natural language understanding unit may determine the structure of the sentence And it is possible to interpret a sentence through analysis of major components, and perform sentence analysis using statistics/analysis, and the like.

In an embodiment, the server 10 may obtain semantic information about the apple by analyzing a sentence including 'apple'. For example, it may be assumed that the word acquired through sentence analysis is “apple.” In this case, the semantic information about the apple may be semantic information indicating one kind of fruit as a noun, but may be used as a verb for another person. It may also be semantic information indicating that a person is saying something wrong, and the server 10 may match semantic information determined to be suitable for a sentence among a plurality of semantic information on 'apple' with 'apple'.

In step S510, the server 10 may obtain an accuracy score for each of the plurality of words based on the matched semantic information and the target patent document.

In this case, the server 10 acquires semantic information about a word acquired through one sentence, but semantic information acquired through this may be inaccurate. Accordingly, the server 10 may acquire the accuracy score of the semantic information matched to the word based on the entire patent document including the sentence.

Accordingly, the server 10 may obtain an accuracy score for the semantic information of a word obtained through sentence analysis based on the semantic information about the same word found in the entire patent document.

For example, the semantic information about 'apple' obtained by analyzing a sentence containing 'apple' is information about an apple representing a fruit, but the semantic information of 'apple' searched in the entire patent document including the sentence is different. In the case of information about telling a person wrong, the server 10 may set the pre-matched semantic information low.

In step S515 , the server 10 may determine a word having an accuracy score equal to or less than a preset value as an erroneous word.

That is, when the accuracy score is less than or equal to a preset value, the server 10 may determine that the semantic information matched with the corresponding word is an erroneous match, and obtain the word as an erroneous word.

Meanwhile, as shown in FIG. 6B , in step S520 , the server 10 may obtain a plurality of semantic information for the erroneous word. Specifically, the server 10 may acquire a plurality of semantic information of the erroneous word from a plurality of patent documents.

In step S525, the server 10 may acquire a plurality of weights for each of the acquired semantic information.

In an embodiment, the server 10 may acquire a weight based on a target patent document including an error word. Specifically, the server 10 may acquire at least one sentence including an erroneous word from all patent documents, and acquire semantic information about the same word as the erroneous word included in the at least one acquired sentence. The server 10 may acquire weights for a plurality of semantic information based on semantic information on at least one word identical to the erroneous word.

In step S530, the server 10 may obtain a weight having the highest degree of association with the target patent document among the plurality of weights.

In step S535, the server 10 may match semantic information corresponding to a weight having the highest degree of relevance with the target patent document to the erroneous word. That is, the server 10 may determine semantic information having the greatest weight among a plurality of semantic information on the erroneous word as semantic information on the erroneous word and correct the erroneous word.

Meanwhile, according to various embodiments of the present invention, it goes without saying that the erroneous word may be a word erroneously parsed due to an error in a morpheme analysis process as well as a case in which incorrect semantic information is matched with a word. In an embodiment, when the sentence for obtaining a word is "Natural language processing using machine learning is performed", the server 10 may obtain machine, learning, use, natural language, and processing as words.

In this case, due to an error in morpheme analysis, words may be acquired such as machine, learning, using eul, natural language, and processing. In this case, the server 10 may determine that 'use E' is an error word and correct it.

Specifically, after obtaining semantic information for each of the acquired words, the server 10 may determine whether the word is an erroneous word based on the degree of association between the semantic information and the target patent document. According to an embodiment, when an acquired word is found in a target patent document including a word having a frequency greater than or equal to a preset frequency, the server 10 may determine that the word is an error-free word. In another embodiment, when the word is found in the entire patent document at a preset frequency or more, the server 10 may determine that the word is not an erroneous word by determining that it is a commonly used word. In another embodiment, when the server 10 does not find semantic information about the word, the server 10 may determine the word as an erroneous word. In another embodiment, when the semantic information matched with the word is different from the target patent document including the word, the server 10 may determine the corresponding word as an erroneous word.

Specifically, the server 10 may acquire a word included in the entire target patent document, match and store semantic information for each of the acquired words. The server 10 may obtain a cluster of a plurality of words having relevant semantic information by clustering the matched words and semantic information. The server 10 may acquire a word cluster in which the number of words included in the cluster is less than or equal to a preset number from among the plurality of word clusters. In this case, the preset number does not mean the number of a plurality of identical words used in the target patent document, but may mean the number of a plurality of words having different shapes. For example, even if the word "use E" is searched for multiple times in the target patent document, since "use E" is the same word, it can be determined as one word. The server 10 may determine the word included in the cluster obtained in the above step as an erroneous word.

However, the present invention is not limited thereto, and of course, the server 10 may determine a word included in a cluster having a distance from another cluster equal to or greater than a preset distance among a plurality of word clusters as an erroneous word. Specifically, the server 10 may obtain the center points of a plurality of clusters, and determine the distance between the clusters based on the obtained center points. For example, the obtained word clusters may be first to fourth word clusters. In this case, the server 10 obtains the distance between the first word cluster and the second word cluster, the distance between the first word cluster and the third word cluster, and the distance between the first word cluster and the fourth word cluster, respectively. , when all three obtained distances are equal to or greater than a preset distance, the first word cluster may be determined as an erroneous word.

Meanwhile, as shown in FIG. 6C , in step S540 , the server 10 may acquire a plurality of words for a sentence included in the target patent document.

Specifically, when a compound noun phrase is formed by combining single words, the meaning may change. Thus, there is a need for the server 10 to obtain a set of compound noun phrases from a single word. Accordingly, the server 10 may obtain a set of compound noun phrases based on a plurality of words.

For example, as a result of analyzing the sentence "Natural language processing using machine learning is performed", the server 10 may obtain words of machine, learning, use, natural language, and processing.

In step S545, the server 10 may obtain a plurality of compound noun phrase set candidates obtained by combining a plurality of words.

For example, the server 10 may obtain complex noun phrase set candidates such as machine learning, learning use, user language, natural language processing, etc. from the obtained combination of words.

In step S550 , the server 10 may obtain a frequency in which the same compound noun phrase set as the obtained plurality of compound noun phrase set candidates is included in the target patent document.

In step S555, the server 10 may determine a compound noun phrase set candidate having a frequency equal to or greater than a preset frequency as the compound noun phrase set.

For example, the server 10 acquires information about which machine learning is found a total of 301 times, learning use is a total of 1 time, user language is a total of 0 times, and natural language processing is found a total of 58 times, and the frequency of appearance is preset. A compound noun phrase candidate set having a frequency greater than or equal to a frequency may be obtained as a compound noun phrase set.

In this case, the compound noun phrase set candidate may include order information and spacing information of words included in the compound noun phrase set candidate.

In an embodiment, the compound noun phrase set is determined as a combination of two or more adjacent words, but in the present invention, a combination of words in which words are spaced apart may also be obtained as a compound noun phrase set. For example, when the server 10 obtains a set of compound noun phrases from a sentence including "a first device among a plurality of devices, a second device among a plurality of devices, and a third device among a plurality of devices", a plurality of A first device of devices, a second device of a plurality of devices, and a third device of a plurality of devices may be obtained as independent compound noun phrase sets, but "(spaced word 1) device of multiple devices" can be obtained as a single compound noun phrase set. Of course, it can also be obtained with In this case, the compound noun phrase set may include spacing information. In the above-described embodiment, the separation information may be information that one word is included between the word “middle” and the word “device”. Of course, according to various embodiments, a plurality of words may be included between words.

개이고, 단어간의 순서가 변경되는 경우 획득되는 복합 명사구 세트의 수는

개이다. 따라서, 문장이 길어지면 길어질수록, 서버(10)는 복합 명사구 세트 검색을 위해 과도한 리소스를 투입하여야 한다. 따라서, 서버(10)는 복합 명사구 세트를 구성하는 단어의 최대 개수를 결정하고, 결정된 개수 내의 복합 명사구 세트를 획득할 수 있다. 이때, 서버(10)는, 단어 획득을 수행하는 대상특허문서와 동일한 기술 분야인 복수개의 특허문서를 획득하고, 획득된 특허문서에 포함된 복합 명사구 세트에 포함된 단어의 최대 개수를 복합 명사구 세트를 구성하는 단어의 최대 개수로 결정할 수 있다. In another embodiment, the server 10 may determine the maximum number of words constituting the compound noun phrase set, and obtain the compound noun phrase set within the determined number. Specifically, when the number of words included in the sentence for obtaining the compound noun phrase set is n and the order between the words is not changed, the number of the compound noun phrase set obtained is

, and the number of sets of compound noun phrases obtained when the order between words is changed is

it's a dog Therefore, as the sentence becomes longer, the server 10 must invest excessive resources for searching the compound noun phrase set. Accordingly, the server 10 may determine the maximum number of words constituting the compound noun phrase set, and obtain the compound noun phrase set within the determined number. At this time, the server 10 acquires a plurality of patent documents that are the same technical field as the target patent document for which word acquisition is performed, and sets the maximum number of words included in the compound noun phrase set included in the obtained patent document to the compound noun phrase set. It can be determined as the maximum number of words constituting

On the other hand, the method for obtaining a set of compound noun phrases according to the present invention is not limited to the above-described method, and of course, it may be obtained by various methods.

In one embodiment, when the plurality of words satisfy a preset condition, the server 10 may change the positions of the plurality of words.

In general, when obtaining a compound noun phrase from a sentence, it is unnecessary to change the order of words included in the sentence. For example, in the sentence of "using machine learning~~", the server 10 needs to obtain a set of complex noun phrases called "machine learning", but acquires a set of complex noun phrases called "learning machine" in which the order of the words is changed. You won't have to. Rather, when acquiring a complex noun phrase set of "treading machine", a complex noun phrase set with a completely different meaning is obtained, and there is a possibility of reducing the performance to be achieved in the present invention.

However, if the preset condition is satisfied, the server 10 may obtain a compound noun phrase set candidate by changing the order of words included in the sentence. In this case, the preset conditions include a condition in which a plurality of words are adjacent, a condition in which a preset sign is included between a plurality of adjacent words, a condition in which a word located to the right of the plurality of adjacent words includes parentheses, and a language in which the plurality of words are different from each other may be a condition.

For example, when the sentence to be analyzed includes "Machine Learning", the server 10 may acquire Machine, Learning, Machine, and Learning as words. Further, the server 10 may acquire "machine learning machine learning" as a set of compound noun phrases. Furthermore, the server 10 may determine that the parentheses, which are preset conditions, are included, and obtain "Machine Learning Machine Learning" as a set of complex noun phrases. That is, the server 10 is expressed in a different language when the compound noun phrase (or word) of "machine learning" and the compound noun phrase (Machine Learning) of another language for the compound noun phrase (or word) exist together, but the same meaning It is possible to obtain two compound noun phrases (or words) having This has the effect of increasing the efficiency of comparison when comparing patent documents written in different languages. That is, a patent document written in Korean is expressed as “Machine Learning”, but a patent document written in English can be expressed as “Machine Learning”, so the server 10 is a composite in both cases. You can increase the efficiency of your search by acquiring it as a set of noun phrases.

Meanwhile, in the above-described embodiment, a condition in which a preset sign includes a condition in which a parenthesis sign is included has been described, but the present invention is not limited thereto. According to various embodiments, the preset sign may be a sign such as a hyphen (-), a slash (/), a tilde (~), or a quotation mark (' or ").

Thereafter, the server 10 may determine a plurality of changed words as a compound noun phrase set candidate.

On the other hand, according to various embodiments of the present invention, when the sentence included in the target patent document has a preset structure, the server 10 excluding the word corresponding to the preset structure among the words included in the sentence is at least one word. can be obtained.

For example, the preset structure may be a standardized structure such as "~ is defined as ~". Specifically, in the target patent document, "~ is defined as ~", "Hereinafter, ~ is referred to as ~." "~ can be understood as ~." A sentence of a structure such as “states ~” may be used to define terms that are not generally used in the technical field. Accordingly, the server 10 may determine a sentence defining a term not generally used in the technical field, and obtain a word defined in the determined sentence.

For example, if the sentence to be analyzed is a sentence "machine learning describes (or defines) OOO", the server 10 determines that the sentence is a preset structure, and words can be obtained.

Thereafter, the server 10 may acquire semantic information of at least one word.

Specifically, when the acquired word of OOO has semantic information commonly used in the technical field, the server 10 may match semantic information corresponding to OOO to OOO. However, when the semantic information is unclear because the OOO term is a term newly defined by the user, the server 10 may match semantic information corresponding to machine learning with the semantic information of OOO.

Thereafter, the server 10 may acquire a word corresponding to the semantic information.

However, the present invention is not limited thereto, and the server 10 may replace the word OOO with a machine learning word when the relevance between machine learning and OOO is greater than or equal to a preset relevance.

On the other hand, according to various embodiments of the present invention, of course, the server 10 may obtain the word except for the template of the patent document.

Specifically, a patent document is a standardized document and has nothing to do with the core content of the invention, but there may be paragraphs describing necessary essential elements in a standardized way. Therefore, when the standardized paragraph is removed and the clue set is obtained, there is an effect of reducing the amount of calculation of the server 10 .

Accordingly, the server 10 may acquire a plurality of identification item information and templates included in the target patent document.

As an embodiment, the template may be set as a paragraph existing at a specific location. For example, the template is the first word in 'Figure 1' in the [Specific Contents for Implementing the Invention] identification item from the sentence included in the line following the [Specific Contents for Implementing the Invention] identification item included in the target patent document. may be the text up to the sentence in which the is searched.

In another embodiment, the template may mean a part common to other patent documents prepared by the creator of the target patent document to be analyzed. That is, the server 10 determines the author of the target patent document, obtains another patent document of the determined author, compares the obtained plurality of patent documents, and determines that the target patent document is similar by more than a preset ratio through the comparison result. Paragraphs or identification items can be obtained as templates. Since a patent document written by the same author has a high probability of having a similar sentence structure, the server 10 may set the criterion for similar judgment in paragraphs or identification items, not in sentence units.

As another embodiment, the server 10 may obtain a text corresponding to a preset identification item as a template. For example, the server 10 may obtain the text included in the identification items such as [simple description of drawings] and [description of symbols] as a template.

In the above-described embodiment, a method of obtaining a template for a general identification item of a patent document for a domestic application has been described, but the present invention is not limited thereto. That is, the server 10 relates to various identification items such as identification items of Korean documents for PCT applications, identification items of English forms for PCT applications, and identification items of forms used in US, Japan, China, and European applications. Of course, you can obtain a template.

Thereafter, the server 10 may acquire a plurality of sentences except for the acquired template.

Thereafter, the server 10 may assign a weight to each of the plurality of identification items.

In one embodiment, when the target patent document has the same format and identification items as in the present specification, the server 10 includes the name of the invention, the technical field, the technology that is the background of the invention, the problem to be solved, the means for solving the problem, The effect of the invention, a brief description of the drawings, specific content for carrying out the invention, description of symbols, claims, summary, representation, and weights for each identification item in the drawings can be given.

In this case, the weight may be assigned by various methods. In one embodiment, the server 10 is a technical field, the background of the invention, the claims, the effect of the invention, the specific content for carrying out the invention, the description of the symbols, the brief description of the drawings, the summary, the problem to be solved Higher weights can be given in the order of , problem solving means, and representativeness.

In another embodiment, the weight is obtained by obtaining the importance score of each word (or compound noun phrase set) included in the target patent document, and after obtaining the importance score for each identification item based on this, the high importance score is applied to the identification item having a high importance score. weights can be assigned. The importance score is an index indicating the importance of a corresponding word (or compound noun phrase set) in a patent document, and the higher the importance score, the higher the word (or compound noun phrase set) may be a keyword of the corresponding patent document.

Meanwhile, the importance score for each identification item may mean a value obtained by dividing the number of words (or compound noun phrase set) after summing up the importance scores of the words (or compound noun phrase sets) included in the identification item. In this case, the word (or compound noun phrase set) used to obtain the importance score for each identification item may be all words (or compound noun phrase set) included in the identification item, but is not limited thereto. For example, the word (or compound noun phrase set) used to obtain the importance score for each identification item is the word (or compound noun phrase set) of the part determined as a template among all the words (or compound noun phrase set) included in the identification item. Of course, it may be a word (or a set of compound noun phrases) except for ).

Thereafter, the server 10 may determine the priority of the identification item to be morphologically analyzed based on the weight of each of the plurality of identification item information.

Thereafter, the server 10 may acquire words for a plurality of sentences from which the template is excluded according to the determined priority.

That is, the server 10 may reduce the amount of calculation by first analyzing an identification item with a high probability of having a keyword, and analyzing an identification item having a relatively low importance based on the analyzed content. Furthermore, when different semantic information for the same word is matched, the server 10 may determine the semantic information obtained from the high-priority identification item as the correct semantic information.

Meanwhile, according to various embodiments of the present invention, of course, the server 10 may obtain a word from the image even when the image is included in the target patent document.

Specifically, when an image is included in the target patent document, the server 10 may determine whether an identification item adjacent to the image exists.

As an embodiment, when an identification item of [Equation] exists in the target patent document and an image exists at the bottom of the identification item, the server 10 may determine the image as an image for the equation. In the case of the equation, since it is highly likely to be the core content of the target patent document, the server 10 may obtain the equation text included in the image.

As another embodiment, when an identification item of [Formula] exists in the target patent document and an image exists at the bottom of the identification item, the server 10 may determine the image as an image for the chemical formula.

Thereafter, when the identification item is adjacent to the image, the server 10 may analyze the image to obtain text included in the image. In this case, the image analysis may be performed by optical character recognition (OCR), but is not limited thereto. For example, if the image is a chemical formula image, of course, the server 10 may obtain a chemical formula corresponding to the image by analyzing the chemical formula structure.

Thereafter, the server 10 may acquire a word based on the acquired text.

7 is a flowchart specifically illustrating a method of calculating an importance score according to an embodiment of the present invention.

In step S610, the server 10, the first appearance ratio of the number of occurrences of the word in the entire patent document compared to the total number of words in the entire patent document, and the word appears in the sentences of the entire patent document compared to the total number of sentences in the entire patent document The first detailed importance level may be calculated based on the second appearance ratio of the number of appeared sentences.

Specifically, the server 10 may count the total number of words in the entire patent document, and may count the number of appearances of the word in the entire patent document.

Thereafter, the server 10 may calculate the number of appearances of words in the entire patent document relative to the total number of words in the entire patent document as the first appearance ratio.

Subsequently, the server 10 may count the total number of sentences in the entire patent document, and may count the number of sentences in which a word appears in the sentences of the entire patent document.

The server 10 may calculate the total number of sentences in the entire patent document compared to the number of sentences in which the word appears among the sentences of the entire patent document as the second appearance ratio.

To this end, the server 10 may search for a word from the entire patent document and search for a sentence including the word. Also, the server 10 may count the total number of words and the total number of sentences by analyzing the sentence components of the entire patent document.

Finally, the server 10 may calculate the first detailed importance based on the first appearance rate and the second appearance rate.

In this case, the server 10 may calculate the first detailed importance level by using Equation 1 below.

<Equation 1>

Here, W1 is the first detailed importance, wpw is the number of occurrences of words in the entire patent document, WPW is the total number of words in the entire patent document, and wps is the number of occurrence sentences in which the word appears in the entire patent document , WPS is the total number of sentences in the entire patent document, and a1 is the adjustment constant of the second appearance ratio.

Referring to Equation 1, the server 10 can calculate the first detailed importance to a greater extent as the number of occurrences of the word in the entire patent document is large and the number of occurrences of the word among the sentences of the entire patent document decreases.

That is, the server 10 may calculate the first detailed importance to a greater extent as a word is repeatedly used in one sentence in the entire patent document.

On the other hand, the server 10 increases the control constant of the second appearance ratio to increase the second appearance ratio even if the number of appearance sentences in which the word appears among the sentences of the entire patent document is small, and reduces the adjustment constant of the second appearance ratio Thus, even if the number of sentences in which a word appears among sentences of the entire patent document is large, the second appearance ratio can be reduced.

Note that, in an embodiment, the type of the detailed importance calculation method is not limited as long as the server 10 calculates the first detailed importance level of the word in the target patent document.

For example, the server 10 may calculate the first detailed importance level by using a Term Frequency-Inverse Document Frequency (TF-IDF) analysis method that calculates importance based on appearance frequency as one of the text analysis methods.

In step S620, the server 10, the third appearance ratio of the number of occurrences of the word in the patent classification information compared to the total number of words in the patent classification information, and the word appears in the sentences of the entire patent document compared to the total number of sentences in the entire patent document The second detailed importance may be calculated based on the fourth appearance ratio of the number of appeared sentences.

Specifically, the server 10 may count the total number of words in the patent classification information, and may count the number of appearances of the word in the patent classification information.

Here, the patent classification information is a code capable of classifying patents according to technical fields, and may be any one of International Patent Classification (IPC), Cooperative Patent Classification (CPC), and F-Term.

Thereafter, the server 10 may calculate the number of appearances of words in the patent classification information compared to the total number of words in the patent classification information as the third appearance ratio.

Subsequently, the server 10 may count the total number of sentences in the patent classification information, and may count the number of sentences in which a word appears in the entire patent document.

The server 10 may calculate the total number of sentences in the entire patent document compared to the number of sentences in which the word appears among the sentences of the entire patent document as the fourth occurrence ratio.

To this end, the server 10 may search for a word from the patent classification information and search for a sentence including the word. Also, the server 10 may count the total number of words and the total number of sentences by analyzing the sentence components of the patent classification information.

Finally, the server 10 may calculate the second detailed importance based on the third and fourth appearance rates.

In this case, the server 10 may calculate the second detailed importance level by using Equation 2 below.

<Equation 2>

Here, W ₂ is the second detailed importance level, ipcw is the number of occurrences of words in the patent classification information, IPCW is the total number of words in the patent classification information, and ipcs is the number of occurrence sentences in which the word appears among all sentences of the patent document. , IPCS is the total number of sentences in the entire patent document, and a2 is the adjustment constant of the fourth appearance ratio.

Referring to Equation 2, the server 10 can calculate the second detailed importance as the number of occurrences of the word in the patent classification information increases and the number of occurrences of the word among the sentences of the entire patent document decreases.

That is, the server 10 may calculate the second detailed importance to a greater extent as a word is repeatedly used in one sentence in the patent classification information.

On the other hand, the server 10 increases the adjustment constant of the fourth appearance rate, so that even if the number of appearance sentences in which a word appears among the sentences of the entire patent document is small, the fourth appearance rate can be increased, and the adjustment constant of the fourth appearance rate is decreased Thus, the fourth appearance ratio can be reduced even if the number of occurrence sentences in which the word appears among the sentences of the entire patent document is large.

Note that, in an embodiment, the type of the detailed importance calculation method is not limited as long as the server 10 calculates the second detailed importance level of the word in the patent document.

For example, the server 10 may calculate the second detailed importance level by using a Term Frequency-Inverse Document Frequency (TF-IDF) analysis method that calculates importance based on appearance frequency as one of the text analysis methods.

In step S630, the server 10 may calculate an influence value of each search patent document based on reference information of each search patent document, and calculate a third detailed importance of the search patent document using the influence value.

Specifically, the server 10 may search for a search patent document including a word that is a target for calculating an importance score among all patent documents.

The server 10 may calculate an influence value of each of the searched patent documents based on the reference information of each of the searched patent documents.

Specifically, the server 10 calculates an influence value based on the number of items, the number of citations and the number of citations, which are reference information, that is, the number of items identical to one or more of the applicant, inventor, and right holder, which are reference information of the search patent document, as the other patent documents. have.

That is, the server 10 may calculate the degree to which the searched patent document is related to another patent document as an influence value. For example, when the searched patent document is cited from several patent documents, the server 10 may determine that the searched patent document has an influence on the corresponding patent document and calculates it as an influence value of the searched patent document.

The server 10 may calculate the third detailed importance of the search patent document by using the calculated influence value for each of the search patent documents.

Specifically, the server 10 may calculate the average of the calculated influence values for each of the search patent documents, and calculate the calculated average as the third detailed importance level.

At this time, the server 10 compares each of the first detailed importance, the second detailed importance, and the third detailed importance with the minimum detailed importance value set corresponding to each of the first detailed importance, the second detailed importance, and the third detailed importance, , a recalculation process may be performed for the detailed importance less than the minimum detailed importance value.

Thereafter, if the first detailed importance, the second detailed importance, and the third detailed importance are equal to or more than the minimum detailed importance value, the server 10 adds up a plurality of the first detailed importance, the second detailed importance, and the third detailed importance to form an importance score. can be calculated.

The server 10 may output the calculated importance score to an internal memory or processor, or may transmit it to an external server.

8A and 8B are flowcharts illustrating a method for obtaining a similar patent document and a method for determining a similar sentence according to an embodiment of the present invention.

Specifically, as shown in FIG. 8A , in step S705 , the server 10 may acquire a plurality of words included in the target patent document.

In step S710 , the server 10 may acquire a plurality of target patent clusters by clustering the plurality of acquired words.

In this case, various clustering techniques may be applied to clustering according to the present invention. In an embodiment, the server 10 may acquire a plurality of target patent clusters using an Elasticsearch clustering clustering technique. In another embodiment, the server 10 uses at least one clustering technique among K-means clustering, DBSCAN (Density-based spatial clustering of applications with noise) clustering, Hierarchical clustering, and a mixed Gaussian clustering technique. A patent cluster can be obtained.

In step S715, the server 10 obtains a plurality of midpoints for each of the plurality of target patent clusters, and based on the plurality of midpoints for each of the plurality of target patent clusters and the number of words included in the target patent clusters The location of the patent document can be determined.

In one embodiment, the center of the target patent cluster may be obtained based on the positions of a plurality of words included in the cluster, and the location of the target patent document may be the center of gravity of the midpoints of the plurality of target patent clusters. have.

In step S720 , the server 10 may acquire a plurality of words included in the patent document, and cluster the acquired words to obtain a plurality of patent clusters.

In step S725, the server 10 acquires a plurality of midpoints for each of the plurality of patent clusters, and the position of the patent document based on the plurality of midpoints for each of the plurality of patent clusters and the number of words included in the patent cluster can be judged

As an embodiment, the midpoint of the patent cluster may be obtained based on the positions of a plurality of words included in the corresponding cluster, and the position of the patent document may be the center of gravity with respect to the midpoints of the plurality of patent clusters.

In step S730, when the position of the target patent document and the position of the patent document are within a preset distance, the server 10 may determine the patent document as a similar patent document.

In one embodiment, the server 10 may determine a plurality of patent documents existing within a preset distance from the location of the target patent document as similar patent documents, and the plurality of similar patent documents having a close distance to the target patent document are determined in the order of the plurality of similar patent documents. Similar patent documents of

Meanwhile, according to various embodiments of the present invention, the server 10 may obtain a plurality of sentences from the target patent document and the similar patent document and sort them in order of importance.

Specifically, the server 10 may acquire a plurality of words for each of a plurality of sentences included in the target patent document and the similar patent document.

Thereafter, the server 10 may obtain a plurality of importance scores for each of the obtained plurality of words. A method of obtaining the importance score will be described later.

Thereafter, the server 10 may determine the importance score of the sentence included in the target patent document based on the plurality of importance scores.

In an embodiment, the server 10 may obtain a degree of association between a plurality of words and the target patent document, and determine the total sum of importance scores of words having the obtained association degree equal to or greater than a preset value as the importance score of the sentence. . In this case, the degree of association may be obtained through various methods. For example, the degree of association between a word and a target patent document may be obtained based on a plurality of patent documents corresponding to the IPC field corresponding to the target patent document. Specifically, the degree of association between a word and a target patent document may be obtained based on the number of times the corresponding word appears in a plurality of patent documents corresponding to the IPC field. As an embodiment, the higher the number of times a corresponding word appears in a plurality of patent documents corresponding to the IPC field, the higher the correlation between the word and the target patent document.

Thereafter, the server 10 may sort the plurality of sentences in order of importance based on the importance score of each of the plurality of sentences.

Furthermore, the server 10 may group a plurality of sentences into n groups. In this case, a method of grouping a plurality of sentences may be various. In an embodiment, the n groups may be grouped to include the same number of sentences. As another embodiment, the n groups may be grouped according to a preset importance score. For convenience of explanation, it is hereinafter assumed that the importance score of the sentences included in the kth group is greater than the importance score of the sentences included in the k+1th group.

The server 10 may acquire the first sentence of the target patent document and the second sentence of the similar patent document based on the importance score. Specifically, the server 10 may classify the plurality of sentences included in the target patent document into n preset groups based on the importance score.

Thereafter, the server 10 may classify the plurality of sentences included in the similar patent document into n preset groups based on the importance score.

Thereafter, when the first sentence is included in a first group among n groups of the target patent document, the server 10 may acquire the second sentence as all sentences included in the similar patent document.

Thereafter, when the first sentence is included in the k-th group among the n groups of the target patent document, the server 10 converts the second sentence to the first to second groups among the n groups of the similar patent document. (n-k+1) Can be obtained by sentences included in the group. In this case, n may be a natural number greater than or equal to 1, and k may be a natural number less than or equal to n.

For example, it may be assumed that the target patent document and the similar patent document are classified into three groups. At this time, when the first sentence is included in the first group, the server 10 compares the first sentence with all sentences included in the first group to the third group (ie, the entire similar patent document) of the similar patent document. evaluation results can be obtained. Also, when the first sentence is included in the second group, the server 10 may obtain an evaluation result by comparing the first sentence with all sentences included in the first group and the second group of the similar patent document. Also, when the first sentence is included in the third group, the server 10 may obtain an evaluation result by comparing the first sentence with all sentences included in the first group of similar patent documents.

That is, the server 10 obtains evaluation results only for sentences determined to be important sentences according to the importance of the sentences, thereby preventing an evaluation result acquisition process for unnecessary or less important sentences, thereby reducing the amount of calculation. Specifically, since the first group of the target patent document is a sentence with high importance, it is compared with all similar patent documents, and since the third group of the target patent document is a sentence with low importance, it is compared with the sentences of the first group with high importance among similar patent documents By comparing, the server 10 can omit the comparison between the third group of (less important) target patent documents and the third group of (less important) similar patent documents.

Meanwhile, according to various embodiments of the present invention, the server 10 may determine whether to group the target patent document and the similar patent document into several groups. Specifically, the server 10 may determine the value of n based on the current occupancy of the GPU of the server 10 and the required accuracy. When the GPU occupancy is high or the required accuracy is low, the server 10 may increase the value of n, and when the GPU occupancy is low or the required accuracy is high, the server 10 may decrease the value of n. For example, it can be assumed that the target patent document and the similar patent document are composed of 6 sentences, and n is 3 and n is 6. When n is 6, the server 10 compares 6+5+4+3+2+1=21 sentences, and when n is 3, the server 10 compares 6+6+4+4+2+ 2 = 24 sentences will be compared. That is, as the value of n increases, the number of sentences to be compared in the target patent document and the similar patent document decreases, and as the value of n decreases, the number of sentences to be compared in the target patent document and the similar patent document increases. ) If the amount of computation (GPU occupancy) that it can currently process is small or the required accuracy is low, the server 10 may increase the value of n.

Meanwhile, in the above-described embodiment, the case in which the target patent document and the similar patent document are grouped in the same number of n has been described, but the present invention is not limited thereto. For example, it goes without saying that the target patent document and the similar patent document may be grouped in different numbers.

Meanwhile, as shown in FIG. 8B , in step S735 , the server 10 may obtain a similarity score and a dissimilarity score of the first sentence and the second sentence, respectively.

Specifically, the similarity score is an index for determining whether the first sentence and the second sentence are related sentences, and the dissimilarity score is an indicator for determining whether the first sentence and the second sentence are unrelated sentences.

In an embodiment, the similarity score and dissimilarity score may be obtained by acquiring semantic information of words included in the first sentence and the second sentence, and determining whether the acquired semantic information is semantic information about the same subject. have.

In another embodiment, the server 10 may determine the sentence structure of the first sentence and the second sentence and use it to calculate the similarity score and the dissimilarity score. For example, the server 10 may increase the similarity score when the sentence structures of the first sentence and the second sentence are similar, and increase the dissimilarity score when the sentence structures of the first sentence and the second sentence are dissimilar. .

In step S740 , when the dissimilarity score is equal to or greater than a preset score, the server 10 may determine that the first sentence and the second sentence are unrelated sentences.

In step S745, when the similarity score is equal to or greater than a preset score, the server 10 may acquire at least one word not included in the second sentence among words included in the first sentence.

That is, even if the first sentence and the second sentence are related sentences, the first sentence and the second sentence are not sentences having the same meaning. Accordingly, the server 10 may determine whether the first sentence and the second sentence are the same sentence or whether a sentence on the same subject or a sentence with a difference exists.

In an embodiment, the server 10 may determine at least one word not included in the second sentence among words included in the first sentence in order to determine the difference between the first sentence and the second sentence.

In another embodiment, the server 10 may determine the sentence structure of the first sentence and the second sentence.

In step S750, the server 10 obtains at least an importance score for each of at least one word not included in the second sentence, and determines whether a word with a preset importance score or higher among the obtained at least one importance score exists. can judge

In step S755 , the server 10 may determine the first sentence and the second sentence as matching sentences when there is no word equal to or greater than the preset importance score.

In step S760 , the server 10 may determine the first sentence and the second sentence as mismatched sentences when there is a word equal to or greater than a preset importance score.

That is, even when at least one word of the first sentence is not included in the second sentence, there may exist a case where a word having a low importance score is not related to the similarity between the first sentence and the second sentence. Accordingly, the server 10 acquires a word when the importance score of at least one word not included in the second sentence is equal to or greater than a preset importance score, and the first sentence and the second sentence match the obtained word based on the acquired word It is possible to determine whether the sentence is a cognitively inconsistent sentence.

In this case, in the above-described step S180, when the first sentence and the second sentence are not related, the server 10 may not include the second sentence in the generated prior art research report. The server 10 may generate a prior art research report based on a second sentence identical to the first sentence or a second sentence similar to the first sentence.

According to an embodiment, the server 10 may obtain a plurality of sentences of the target patent document, sort the sentences in order of importance, and determine whether there are similar sentences in the similar patent document from the highest importance sentence. For example, the server 10 may obtain sentences 1-1 to 1-n th sentences by arranging the target patent documents in order of importance.

The server 10 may obtain the second sentence of the similar patent document for the 1-1 sentence by comparing the 1-1 sentence with the highest importance with the similar patent document. In the same way, the server 10 compares the sentence 1-2 with the similar patent document to obtain the second sentence of the similar patent document for the sentence 1-2, and compares the sentence 1-n with the similar patent document. Thus, the second sentence of the similar patent document for the 1-nth sentence can be obtained.

The server 10 may select only a sentence related to the first sentence from among the plurality of acquired second sentences. According to various embodiments, the server 10 may display the second sentence for all sentences (ie, the 1-1 sentence to the 1-nth sentence) of the target patent document, but is not limited thereto. It goes without saying that only the second sentence of a preset high importance sentence (eg, a sentence corresponding to the top 30% as a result of sorting in order of importance) among all sentences of the patent document may be displayed.

Meanwhile, according to various embodiments of the present invention, the server 10 includes the target patent document, the claims of at least one similar patent document for the target patent document, and the claims of the target patent document and at least one similar patent document. It can be judged whether a specific sentence of the target patent document infringes on a similar patent document based on the importance score for the word used.

9 is a flowchart illustrating a specific method of generating a thesaurus according to an embodiment of the present invention.

In step S810, the server 10 may convert a word into a word vector through Word2Vec learning based on the patent document.

In step S820 , the server 10 may calculate a distance between any two word vectors among the word vectors and calculate the calculated distance as a degree of similarity.

In step S830, the server 10 checks whether the similarity between any two word vectors among the word vectors is less than a preset reference similarity, and if the similarity between any two word vectors among the word vectors is less than the preset reference similarity, the two words Two words corresponding to the vector may be grouped into a similar word group.

In step S840 , the server 10 may check whether any one patent document among all the acquired patent documents satisfies the noise document condition, and remove the patent document that satisfies the noise document condition.

Meanwhile, according to various embodiments of the present disclosure, the server 10 may select a reference word vector from among word vectors.

Here, the reference word vector may be a word vector corresponding to a reference word of a similar word group to be described later. For example, if the reference word is an apple, words included in the analogous word of the apple may be included in the analogous word group.

Thereafter, the server 10 may calculate a similarity between the reference word vector and another word vector. In this case, the server 10 may calculate the similarity between the reference word vector and another word vector in the same way as the above-described similarity calculation method.

Subsequently, the server 10 may determine whether a word corresponding to the other word vector is included in the similarity group based on the reference word corresponding to the reference word vector based on the similarity between the reference word vector and the other word vector.

Specifically, if the similarity between the reference word vector and another word vector is less than the preset reference similarity, the server 10 determines that the word corresponding to the other word vector is included in the similarity group based on the reference word corresponding to the reference word vector. can be checked

Thereafter, when it is confirmed that the word corresponding to the other word vector is included in the similarity group based on the reference word corresponding to the reference word vector, the server 10 determines the similarity of the word corresponding to the other word vector based on the reference word. can be included in the group.

10 is an exemplary diagram for explaining a process of deriving an evaluation result using a first artificial intelligence model and a second artificial intelligence model according to an embodiment of the present invention.

As shown in FIG. 10 , the first artificial intelligence model may receive a target patent document as an input value and obtain a similar patent document for the target patent document. In this case, there may be a plurality of similar patent documents for the target patent document, and if there are a plurality of similar patent documents, the server 10 may sort the plurality of similar patent documents based on the association with the target patent document. of course there is

The first database is configured to store the entire patent document. Specifically, the first database obtains the words of each of the entire patent documents through morphological analysis of each of the entire patent documents, obtains the thesaurus for the obtained words, and obtains the importance score for each of the obtained words. , the obtained importance score may be matched to a word and stored.

The first artificial intelligence model may acquire a similar patent document corresponding to the target patent document among all patent documents stored in the first database.

Meanwhile, when the target patent document and the similar patent document are obtained through the first artificial intelligence model, the server 10 may extract a first sentence of the target patent document and extract a second sentence of the similar patent document. The first sentence and the second sentence may be obtained through the various methods described above.

Furthermore, the second artificial intelligence model may obtain the evaluation results for the first sentence and the second sentence by inputting the first sentence and the second sentence as input values. In this case, the evaluation result may be a result of whether the first sentence and the second sentence are a matching sentence, a discordant sentence, or an unrelated sentence, and may be obtained according to various embodiments described above.

The second database may store a patent document and a prior search report matched to the patent document. In this case, the prior search report may include information on a similar patent document determined to be similar to a patent document and a specific sentence of a similar patent document corresponding to a specific sentence of the patent document.

11 is a block diagram of an apparatus according to an embodiment of the present invention.

The processor 102 may include one or more cores (not shown) and a graphic processing unit (not shown) and/or a connection path (eg, a bus, etc.) for transmitting and receiving signals to and from other components. .

The processor 102 according to an embodiment performs the method described with reference to FIGS. 1 to 9 by executing one or more instructions stored in the memory 104 .

For example, the processor 102 obtains at least one word based on the target patent document by executing one or more instructions stored in the memory, obtains an importance score of the obtained at least one word, and generates the target patent document 1 input into the artificial intelligence model to obtain a similar patent document, obtain a plurality of sentences included in the target patent document, obtain a plurality of sentences included in the similar patent document, and Obtaining the first sentence, acquiring the second sentence among the plurality of sentences included in the similar patent document, and inputting the first sentence and the second sentence into the second artificial intelligence model, An evaluation result may be obtained, and a prior art search report for the target patent document may be generated based on the evaluation result.

On the other hand, the processor 102 is a RAM (Random Access Memory, not shown) and ROM (Read-Only Memory: ROM) for temporarily and / or permanently storing a signal (or data) processed inside the processor 102. , not shown) may be further included. In addition, the processor 102 may be implemented in the form of a system on chip (SoC) including at least one of a graphic processing unit, a RAM, and a ROM.

The memory 104 may store programs (one or more instructions) for processing and controlling the processor 102 . Programs stored in the memory 104 may be divided into a plurality of modules according to functions.

The steps of a method or algorithm described in relation to an embodiment of the present invention may be implemented directly in hardware, as a software module executed by hardware, or by a combination thereof. A software module may contain random access memory (RAM), read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, hard disk, removable disk, CD-ROM, or It may reside in any type of computer-readable recording medium well known in the art to which the present invention pertains. The components of the present invention may be implemented as a program (or application) to be executed in combination with a computer, which is hardware, and stored in a medium. The components of the present invention may be implemented as software programming or software components, and similarly, embodiments may include various algorithms implemented as data structures, processes, routines, or combinations of other programming constructs, including C, C++ , Java, assembler, etc. may be implemented in a programming or scripting language. Functional aspects may be implemented in an algorithm running on one or more processors.

As mentioned above, although embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art to which the present invention pertains can realize that the present invention can be embodied in other specific forms without changing its technical spirit or essential features. you will be able to understand Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

10 : Server
20: user terminal

Claims (1)

In the control method of the similarity determination system of patent documents,
Step, by the server, obtaining a target patent document;
obtaining, by the server, at least one word based on the target patent document;
obtaining, by the server, an importance score of the obtained at least one word;
obtaining, by the server, the similar patent document by inputting the target patent document into a first artificial intelligence model;
obtaining, by the server, a plurality of sentences included in the target patent document, and acquiring a plurality of sentences included in the similar patent document;
obtaining, by the server, a first sentence among a plurality of sentences included in the target patent document, and acquiring a second sentence among a plurality of sentences included in the similar patent document;
inputting, by the server, the first sentence and the second sentence into a second artificial intelligence model, and obtaining evaluation results for the first sentence and the second sentence; and
and generating, by the server, a prior art search report for the target patent document based on the evaluation result,
The step of obtaining the importance score comprises:
obtaining, by the server, a word to be calculated an importance score from the target patent document;
The server includes the first detailed importance of the word in the entire patent document, the second detailed importance of the word in the patent classification information corresponding to the technical field information of the target patent document, and the word in the entire patent document calculating at least one detailed importance among third detailed importance levels of the searched patent document; and
and calculating, by the server, the importance score of the word based on at least one of the first detailed importance level, the second detailed importance level, and the third detailed importance level.

Images