A key information analysis system for technical patent data

Abstract

The application provides a key information analysis system for technical patent data, and relates to the technical field of patent analysis, comprising a collection and storage layer, an analysis layer and an application layer, wherein the collection and storage layer comprises a collection module and a storage module, the analysis layer comprises a text conversion and extraction module, a keyword benchmark library, an extraction module, a specific gravity value analysis module, a summary module and a classification module; the collection module is used for collecting technical patent data in all formats, and the storage module is used for storing the technical patent data; after collecting the technical patent data, the application takes the keyword benchmark library as a benchmark, extracts relevant and similar keywords and their contexts in the patent text data, summarizes the extracted keywords and their contexts into texts representing the fields, advantages and technical means of the patent technology, and induces them into short data reports, so that a person can quickly connect the key information of each patent from the short data reports, save understanding time and facilitate screening.

Description

Technical Field

[0001] This invention relates to the field of patent analysis technology, and in particular to a key information analysis system for technical patent data. Background Technology

[0002] A patent, literally, refers to exclusive rights and interests. A patent means a public letter or document. In modern times, a patent is generally a document issued by a government agency or a regional organization representing several countries based on an application. This document records the content of the invention or creation and creates a legal status for a certain period of time, that is, under normal circumstances, others can only implement the patented invention or creation with the permission of the patentee. The patent exists in the form of technical patent data.

[0003] When enterprises search for and retrieve a batch of patent data, they need to read a large amount of information. A single patent document is usually quite long, and understanding it requires a lot of time for analysis. When searching for a target patent among multiple patents, it is necessary to read and understand it sequentially, which also takes a lot of time. Moreover, some patent content is verbose and complex, making it difficult to fully understand when reading and understanding large quantities of data, which affects the subsequent development and application of the patent. Therefore, this invention proposes a key information analysis system for technical patent data to solve the problems existing in the prior art. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a key information analysis system for technical patent data. This system extracts keywords and their contexts, summarizing them into text representing the patent's field, advantages, and technical means, and then compiling them into short data reports. Users can quickly connect to the key information of each patent from these short data reports, saving time and facilitating filtering.

[0005] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a key information analysis system for technical patent data, comprising a data acquisition and storage layer, an analysis layer, and an application layer. The data acquisition and storage layer includes a data acquisition module and a storage module. The analysis layer includes a text conversion and extraction module, a keyword benchmark library, an extraction module, a weighting value analysis module, a summary and induction module, and a classification module.

[0006] The acquisition module is used to acquire technical patent data in all formats. The storage module is used to store the technical patent data. The text conversion and extraction module is used to convert all formats of technical patent data into patent text data. The keyword benchmark library includes all words related to and similar to the core key content of the patent data. The extraction module is used to extract relevant and similar keywords and their context from the patent text data based on the keyword benchmark library. The summary and generalization module is used to summarize the extracted keywords and their context into text representing the field, advantages, and technical means of the patent technology, and summarize them into a short data report. The weighting analysis module is used to calculate the frequency of each keyword in the corresponding patent text data, generate the focus of the patent technology, and summarize it synchronously in the short data report. The classification module is used to classify and format the short data analyzed from multiple patent text data.

[0007] Further improvements include: the acquisition module is used to receive and transmit XML, text, PDF, images and other files of any format, and integrates Big data technology, using distributed fast exchange technology for information transmission.

[0008] A further improvement is that the storage module includes a storage library and a retrieval module. The storage library is used to store all data collected by the acquisition module and to add timestamps. The retrieval module provides a retrieval function to retrieve technical patent data from the storage library based on the timestamps.

[0009] Further improvements are made in that: the text conversion and extraction module includes a text extraction module and an image conversion module. The text extraction module is used to extract the text of technical patent data in all document formats and convert it into an editable text document. The image conversion module obtains the visual features of the image through CNN, obtains the sequence features of the image through RNN, obtains the text sequence information through classifier CTC or decoder attention, extracts the document from the image based on OpenCV, and then converts it into an editable text document.

[0010] Further improvements are made in the following ways: The keyword benchmark library includes a background technology benchmark library, a beneficial effect benchmark library, a working principle benchmark library, and a domain benchmark library. The background technology benchmark library includes the following keywords: "background," "existing technology," "disadvantages," "poor," and words similar to or close to the above keywords. The beneficial effect benchmark library includes the following keywords: "effect," "efficiency," "excellent," "improvement," "promotion," "good," and words similar to or close to the above keywords. The working principle benchmark library includes the following keywords: "through," "drive," "use," and words similar to or close to the above keywords. The domain benchmark library includes the following keywords: "technology" and "domain."

[0011] Further improvements are made in that: the extraction module includes a keyword scanning and recognition module and a context acquisition module. The keyword scanning and recognition module scans the patent text data based on a keyword benchmark library and uses contrastive neural network technology to mark the keywords in the patent text data. The context acquisition module is used to extract 1-3 sentences belonging to the context of the keyword based on the keyword marking and integrate the extraction set.

[0012] Further improvements are made in the following ways: The summarization module includes a semantic analysis module, a deduplication module, and a report generation module. The semantic analysis module is used to access the extraction set and analyze the semantics of the statements in the extraction set using the DINFO-OEC unstructured model combined with natural language processing (NLP) technology. It analyzes the field, advantages, and technical means of the patent technology represented by each statement. The deduplication module is used to delete duplicate sentences based on the analyzed semantics and retain at least one sentence that represents the field, advantages, and technical means. The report generation module is used to combine the deduplicated sentences into short data representing the corresponding technology patents and to separate and integrate the short data of multiple technology patents into a single report.

[0013] A further improvement is that the weighting analysis module is based on the TF-IDF statistical document retrieval algorithm to evaluate the frequency of each keyword in the extracted set, thereby determining the importance of the keyword to the technology patent, and using the keyword as a representative word of the patent technology focus, and summarizing it into the short data of the corresponding technology patent.

[0014] Further improvements are made in the following aspects: The classification module includes a categorization system and a binding system. The categorization system identifies short data of multiple technology patents in the report and connects to the DINFO-OEC unstructured model combined with natural language processing (NLP) technology to analyze semantics. It categorizes the short data of different technology patents based on the domain, and within the same domain, it further categorizes the short data of different technology patents based on the technology direction. The data is then formatted on the report. The binding system binds each short data entry to the original data of the technology patent, providing a function for querying the original data.

[0015] Further improvements are made in that: the application layer includes a display module and a security module. The display module is used to display the reports generated by the summary module on the human-computer interaction panel. The security module is used to perform encryption verification on the entire system using methods such as user access authorization, user access detection, user control authorization, data export authorization, reverse control authorization, and data control encryption.

[0016] The beneficial effects of this invention are as follows:

[0017] 1. After collecting technical patent data, this invention uses a keyword benchmark library as a benchmark to extract relevant and similar keywords and their context from the patent text data. By summarizing the extracted keywords and their context into text representing the field, advantages, and technical means of the patent technology, and compiling them into short data reports, users can quickly connect to the key information of each patent from the short data reports, saving understanding time and facilitating filtering.

[0018] 2. This invention scans and identifies patent data and collects context based on keywords from background technology benchmark libraries, beneficial effect benchmark libraries, working principle benchmark libraries, and domain benchmark libraries. Key sentences are extracted and integrated into an extraction set. The semantics of the sentences in the extraction set are analyzed using the DINFO-OEC unstructured model combined with natural language processing (NLP) technology. This analysis reveals the domain, advantages, and technical means of the patent technology represented by each sentence. As a result, the generated short data reports can accurately represent the key information of the patent data, making them easier to understand and facilitating subsequent development and application of the patent.

[0019] 3. This invention uses a weighting analysis module to evaluate the frequency of each keyword in the extracted set, thereby determining the importance of the keyword to the technology patent. This keyword is then used as a representative word of the patent's technical focus. When a patent has multiple technical points, the focus of the focus can be determined, thus diversifying the functions.

[0020] 4. This invention categorizes short data of different technology patents based on their respective fields using a classification module. Within the same field, it further categorizes short data of different technology patents based on their technical direction, thus arranging the data for easy reference. Each short data entry is also linked to the original data of the corresponding technology patent, facilitating rapid retrieval of the original data and improving efficiency. Attached Figure Description

[0021] Figure 1 This is a diagram illustrating the composition of the present invention. Detailed Implementation

[0022] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0023] Example 1

[0024] according to Figure 1 As shown in the figure, this embodiment proposes a key information analysis system for technical patent data, including a collection and storage layer, an analysis layer, and an application layer. The collection and storage layer includes a collection module and a storage module, and the analysis layer includes a text conversion and extraction module, a keyword benchmark library, an extraction module, a weight value analysis module, a summary and induction module, and a classification module.

[0025] The acquisition module is used to collect technical patent data in all formats. The storage module is used to store the technical patent data. The text conversion and extraction module is used to convert all formats of technical patent data into patent text data. The keyword benchmark library includes all words related to and similar to the core key content of the patent data. The extraction module is used to extract relevant and similar keywords and their context from the patent text data based on the keyword benchmark library. The summarization module is used to summarize the extracted keywords and their context into text representing the patent technology's field, advantages, and technical means, and summarize it into a short data report. The weighting analysis module is used to calculate the frequency of each keyword in the corresponding patent text data, generate the patent technology's focus, and summarize it synchronously into the short data report. The classification module is used to classify and format the short data analyzed from multiple patent text data. After collecting technical patent data, this invention uses the keyword benchmark library as a benchmark to extract relevant and similar keywords and their context from the patent text data. By summarizing the extracted keywords and their context into text representing the patent technology's field, advantages, and technical means, and summarizing it into a short data report, the reader can quickly connect to the key information of each patent from the short data report, saving understanding time and facilitating filtering.

[0026] The acquisition module is used to receive and transmit files in XML, text, PDF, images, and any other format. It integrates Big Data technology and employs distributed high-speed exchange technology for information transmission, making it easier to acquire and utilize files in various formats.

[0027] The storage module includes a database and a retrieval module. The database stores all data collected by the acquisition module and timestamps it. The retrieval module provides a search function to retrieve technical patent data from the database based on the timestamp. The advantage is that it allows for searching by time, making it convenient to consult.

[0028] The text conversion and extraction module includes a text extraction module and an image conversion module. The text extraction module is used to extract the text of technical patent data in all document formats and convert it into an editable text document. The image conversion module obtains the visual features of the image through CNN, obtains the sequence features of the image through RNN, obtains the text sequence information through classifier CTC or decoder attention, extracts the document from the image based on OpenCV, and then converts it into an editable text document.

[0029] OpenCV, in conjunction with the TEXT extension module, performs text recognition. OpenCV is based on extreme region text localization and recognition, and adds a convolutional neural network to achieve text detection. This dual detection and recognition makes the extraction of documents from images more accurate.

[0030] The keyword benchmark library includes a background technology benchmark library, a beneficial effect benchmark library, a working principle benchmark library, and a field benchmark library. The background technology benchmark library includes keywords such as "background," "existing technology," "disadvantages," "poor," and similar or related terms. The beneficial effect benchmark library includes keywords such as "effect," "efficiency," "excellent," "improvement," "promotion," "good," and similar or related terms. The working principle benchmark library includes keywords such as "through," "drive," "use," and similar or related terms. The field benchmark library includes keywords such as "technology" and "field." This comprehensive library, encompassing background technology, beneficial effects, working principles, and fields, represents the key technical information of the patent.

[0031] The extraction module includes a keyword scanning and recognition module and a context acquisition module. The keyword scanning and recognition module scans the patent text data based on a keyword benchmark library and uses contrastive neural network technology to identify keywords within the patent text data. The context acquisition module extracts 1-3 sentences belonging to the context of each keyword based on its identification, and integrates these extraction sets. During comparison, a YOYO neural network is used, employing convolutional neural networks as the basic framework on the TensorFlow and PyTorch platforms for cross-comparison verification, quickly determining the comparison results and identifying keywords in the patent text data.

[0032] The summary and induction module includes a semantic analysis module, a deduplication module, and a report generation module. The semantic analysis module accesses the extraction set and analyzes the semantics of the statements in the extraction set using the DINFO-OEC unstructured model combined with natural language processing (NLP) technology. It analyzes the field, advantages, and technical means of the patent technology represented by each statement. The deduplication module is used to delete duplicate sentences based on the analyzed semantics, retaining at least one sentence that represents the field, advantages, and technical means. The report generation module is used to combine the deduplicated sentences into short data representing the corresponding technology patents and to separate and integrate the short data of multiple technology patents into a single report. This invention scans and identifies patent data and collects context based on keywords from background technology benchmark libraries, beneficial effect benchmark libraries, working principle benchmark libraries, and domain benchmark libraries. Key sentences are extracted and integrated into an extraction set. The semantics of the sentences in the extraction set are analyzed using the DINFO-OEC unstructured model combined with Natural Language Processing (NLP) technology. This analysis reveals the domain, advantages, and technical means of the patent technology represented by each sentence. Consequently, the generated short data reports accurately represent the key information of the patent data, facilitating understanding and providing convenience for subsequent development and application of the patent.

[0033] Natural Language Processing Natural Language Processing (NLP) technology is a collective term for all technologies related to computer processing of natural language. Its purpose is to enable computers to understand and accept instructions input by humans in natural language, and to perform translation functions from one language to another. The DINFO-OEC platform supports a three-dimensional, multi-dimensional business modeling capability, combining natural language processing, deep learning, and other statistical text mining algorithms. Based on the platform's three-dimensional business model and intelligent semantic perception technology, it provides intelligent understanding and automated processing capabilities for unstructured big data, enabling multi-dimensional business tagging of text knowledge and converting disordered unstructured information into structured data that meets business needs. The DINFO-OEC platform supports integration with mainstream big data platforms such as Hadoop and Spark. Utilizing the distributed storage and Map / Reduce distributed computing capabilities provided by the Hadoop platform, it enables complex, batch big data analysis and mining. Leveraging the real-time distributed computing capabilities provided by Spark and Kafka, it provides real-time analysis and computing capabilities for massive amounts of data. Integrating mainstream search engine technologies, it supports interactive search functions based on massive historical data. The DINFO-OEC platform supports integration with commonly used intelligent systems, enabling the fusion analysis and mining of structured and unstructured data to maximize the semantic value of big data.

[0034] The weighting analysis module, based on the TF-IDF statistical document retrieval algorithm, evaluates the frequency of each keyword in the extracted set, thereby determining the importance of that keyword to the technology patent. This keyword is then used as a representative term of the patent's technological focus and summarized into the short data of the corresponding technology patent. This invention uses the weighting analysis module to evaluate the frequency of each keyword in the extracted set, thereby determining the importance of that keyword to the technology patent. This allows for the identification of the emphasized technological point when the patent has multiple technical points, providing diversified functionality.

[0035] TF-IDF is a weighted technique used for information retrieval and data mining. TF stands for Term Frequency and IDF stands for Inverse Text Frequency Index. It is used to evaluate the importance of a word to a set of documents. The importance of a word increases proportionally to the number of times it appears in the document, but decreases inversely proportionally to the frequency of its appearance in the corpus.

[0036] The classification module includes a categorization system and a binding system. The categorization system identifies short data entries for multiple technology patents in the report and uses the DINFO-OEC unstructured model combined with Natural Language Processing (NLP) technology to analyze semantics. It categorizes short data entries for different technology patents based on their domain, and further categorizes them based on their technical direction within the same domain. This categorization is then used for layout on the report. The binding system links each short data entry to the original data of that technology patent, providing a function for querying the original data. This invention categorizes short data entries for different technology patents based on their domain, and further categorizes them based on their technical direction within the same domain, thus facilitating easy browsing. Binding each short data entry to the original data of that technology patent also facilitates rapid retrieval of the original data, improving efficiency.

[0037] Example 2

[0038] according to Figure 1 As shown in the figure, this embodiment proposes a key information analysis system for technical patent data, including a collection and storage layer, an analysis layer, and an application layer. The collection and storage layer includes a collection module and a storage module, and the analysis layer includes a text conversion and extraction module, a keyword benchmark library, an extraction module, a weight value analysis module, a summary and induction module, and a classification module.

[0039] The acquisition module is used to collect technical patent data in all formats. The storage module is used to store the technical patent data. The text conversion and extraction module is used to convert all formats of technical patent data into patent text data. The keyword benchmark library includes all words related to and similar to the core key content of the patent data. The extraction module is used to extract relevant and similar keywords and their context from the patent text data based on the keyword benchmark library. The summarization module is used to summarize the extracted keywords and their context into text representing the patent technology's field, advantages, and technical means, and summarize it into a short data report. The weighting analysis module is used to calculate the frequency of each keyword in the corresponding patent text data, generate the patent technology's focus, and summarize it synchronously into the short data report. The classification module is used to classify and format the short data analyzed from multiple patent text data. After collecting technical patent data, this invention uses the keyword benchmark library as a benchmark to extract relevant and similar keywords and their context from the patent text data. By summarizing the extracted keywords and their context into text representing the patent technology's field, advantages, and technical means, and summarizing it into a short data report, the reader can quickly connect to the key information of each patent from the short data report, saving understanding time and facilitating filtering.

[0040] The application layer includes a display module and a security module. The display module displays the reports generated by the summary module on a human-computer interaction panel. The security module employs user access authorization, user access detection, user control authorization, data export authorization, reverse control authorization, and data control encryption to encrypt and verify the entire system. This invention displays the reports generated by the summary module on a human-computer interaction panel, facilitating human-computer operation and review. By setting user access authorization, user access detection, user control authorization, data export authorization, reverse control authorization, and data control encryption, the security of patent data is improved, preventing unauthorized misuse.

[0041] This key information analysis system for patent data collects patent data and, using a keyword benchmark library as a basis, extracts relevant and similar keywords and their context from the patent text data. By summarizing the extracted keywords and their context into text representing the patent technology's domain, advantages, and technical means, it generates short data reports. Users can quickly connect to the key information of each patent from these reports, saving time and facilitating filtering. Furthermore, this invention scans and identifies patent data and collects context based on keywords from background technology benchmark libraries, beneficial effects benchmark libraries, working principle benchmark libraries, and domain benchmark libraries. It extracts key statements, integrates them into an extraction set, and analyzes the semantics of the statements in the extraction set using the DINFO-OEC unstructured model combined with Natural Language Processing (NLP) technology. This analysis reveals the patent technology's domain, advantages, and technical means represented by each statement. Therefore, the generated short data reports accurately represent the key information of the patent data, facilitating understanding and providing convenience for subsequent development and application of the patent. Meanwhile, this invention uses a weighting analysis module to evaluate the frequency of each keyword in the extracted set, thereby determining the importance of the keyword to the technology patent. This keyword is then used as a representative term for the patent's technological focus. When a patent has multiple technical points, it identifies the emphasized technical point, offering diverse functionalities. Furthermore, this invention uses a classification module to categorize short data from different technology patents based on their respective fields. Within the same field, it further categorizes short data from different technology patents based on their technical direction, providing convenient layout for easy browsing. Each short data entry is also linked to the original data of the technology patent, facilitating rapid retrieval of the original data and improving efficiency.

[0042] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A key information analysis system of technical patent data, comprising a collection storage layer, an analysis layer and an application layer, characterized in that: The collection storage layer comprises a collection module and a storage module, and the analysis layer comprises a character conversion extraction module, a keyword benchmark library, an extraction module, a specific gravity value analysis module, a summary and induction module, and a classification module; The collection module is used for collecting technical patent data in all formats, the storage module is used for storing the technical patent data, the character conversion extraction module is used for converting the technical patent data in all formats into patent character data, the keyword benchmark library comprises all words related to the core key content of the patent data, the extraction module is used for extracting relevant and similar keywords and their contexts in the patent character data based on the keyword benchmark library as a benchmark, the summary and induction module is used for summarizing the extracted keywords and their contexts into characters representing the field, advantages, and technical means of the patent technology, and inducing them into a short data report, and the specific gravity analysis module is used for calculating the frequency of each keyword in the corresponding patent character data, generating the focus of the patent technology, and synchronously summarizing into the short data report, and the classification module is used for classifying and typesetting the short data analyzed from multiple patent character data; The extraction module comprises a keyword scanning and identifying module and a context collection module, the keyword scanning and identifying module scans the patent character data based on the keyword benchmark library and a comparison neuron technology, and marks the keywords in the patent character data, and the context collection module is used for extracting 1-3 sentences belonging to the context of the keywords according to the marking of the keywords, and integrating the extracted collection; The summary and induction module comprises a semantic analysis module, a duplicate removal module, and a report generation module, the semantic analysis module is used for accessing the extracted collection, analyzing the semantics of the sentences in the extracted collection by a DINFO-OEC unstructured model combined with a natural language processing NLP technology, and analyzing the field, advantages, and technical means of the patent technology represented by each sentence, the duplicate removal module is used for deleting the repeated sentences according to the analyzed semantics, and retaining at least one sentence representing the field, advantages, and technical means, and the report generation module is used for combining the sentences after the duplicate removal into short data representing the corresponding technical patent, and separating and integrating the short data of multiple technical patents on a report; The specific gravity analysis module evaluates the frequency of each keyword in the extracted collection based on a TF-IDF statistical document retrieval algorithm, so as to judge the importance of the keyword for the technical patent, and takes the keyword as a representative word of the focus of the patent technology, and induces it into the short data of the corresponding technical patent; The classification module comprises a classification system and a binding system, the classification system identifies the short data of multiple technical patents in the report, accesses the DINFO-OEC unstructured model combined with the natural language processing NLP technology to analyze the semantics, classifies the short data of different technical patents based on the field, classifies the short data of different technical patents based on the technical direction under the same field, and typesets the classification on the report, and the binding system binds each short data with the original data of the technical patent, and provides an original data query function.

2. The key information analysis system for technical patent data according to claim 1, characterized in that: The collection module is used for receiving and transmitting XML, text, PDF, image and other formats of files, and is combined with Big data technology, and uses distributed fast exchange technology for information transmission.

3. The key information analysis system for technical patent data according to claim 2, characterized in that: The storage module comprises a storage library and a retrieval module, the storage library is used for storing all data collected by the collection module and is time-stamped, and the retrieval module provides a retrieval function for retrieving technical patent data in the storage library according to the time stamp.

4. The key information analysis system for technical patent data according to claim 3, characterized in that: The text conversion extraction module comprises a text extraction module and a picture conversion module, the text extraction module is used for extracting the text of technical patent data in all document formats and converting the text into an editable text document, the picture conversion module obtains the visual features of an image through CNN, obtains the sequence features of the image through RNN, obtains the text sequence information through a classifier CTC or a decoder attention, extracts the document in the picture based on opencv, and then converts the document into an editable text document.

5. The key information analysis system of technical patent data according to claim 4, characterized in that: The keyword benchmark library comprises a background technology benchmark library, a beneficial effect benchmark library, a working principle benchmark library and a field benchmark library, the background technology benchmark library comprises the following keywords: "background", "prior art", "disadvantage", "poor" and similar and close words to the above keywords, the beneficial effect benchmark library comprises the following keywords: "effect", "efficiency", "excellent", "perfect", "promote", "good" and similar and close words to the above keywords, the working principle benchmark library comprises the following keywords: "through", "drive", "use" and similar and close words to the above keywords, and the field benchmark library comprises the following keywords: "technology", "field".

6. The system for analyzing key information of technical patent data according to any one of claims 1-5, wherein: The application layer comprises a display module and a security module, the display module is used for displaying the report generated by the summary module on a human-computer interaction panel, and the security module uses user access authorization, user access detection, user control authorization, data export authorization, reverse control authorization and data control encryption means to encrypt and verify the whole system.

Images