Intellectual property information analysis and matching system based on big data

Abstract

The invention relates to the technical field of data processing, and discloses an intellectual property information analysis and matching system based on big data, comprising: an acquisition unit acquiring user demands and a historical demand set, performing semantic analysis on the user demands to extract demand keywords, and determining importance; sequencing the demand keywords according to the importance to form a current keyword group; the judgment unit is used for comparing the keyword group with a historical demand set and judging whether a historical matching catalog is taken or not; when the historical matching catalogue is judged to be used, determining a current matching catalogue according to the historical matching catalogue and the current keyword group; when it is judged that the historical matching catalogue is not used, determining a current matching catalogue according to the current keyword group; the processing unit performs clustering analysis on the matching catalog to determine a matching display result; and the storage unit stores the current keyword group and the matched display result. According to the invention, the matching efficiency and accuracy are improved, and the problems of low efficiency, low accuracy and the like in traditional intellectual property matching are solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an intellectual property information analysis and matching system based on big data. Background Art

[0002] Most of the existing intellectual property information matching methods rely on subjective analysis and judgment by personnel, which has the disadvantages of low efficiency and low accuracy, and it is difficult to meet the actual needs of enterprises. For example: most of the current analysis and matching methods only extract the main content of the demand based on keywords, and cannot deeply analyze the semantic association of words. Simple extraction methods may lead to deviations in the understanding of user needs, especially in the presence of polysemous words or industry-specific terms. When matching intellectual property needs, historical demand data is often not effectively utilized, resulting in the system being inefficient when processing a large number of repetitive needs, and unable to quickly provide matching results based on similar historical needs.

[0003] Therefore, it is necessary to design an intellectual property information analysis and matching system based on big data to solve the problems existing in current technology. Summary of the invention

[0004] In view of this, the present invention proposes an intellectual property information analysis and matching system based on big data, aiming to solve the problems existing in the current intellectual property information matching technology that relies on subjective analysis and judgment by personnel, and has the disadvantages of low efficiency and low accuracy.

[0005] The present invention proposes an intellectual property information analysis and matching system based on big data, comprising: A collection unit is configured to collect user needs and historical demand sets, perform semantic analysis on the user needs, extract demand keywords, determine the importance of the demand keywords according to the number of occurrences of each demand keyword and the keyword coverage; sort all the demand keywords according to the importance to form a current keyword group; A judgment unit is configured to compare the keyword group with each historical keyword group in the historical demand set, and judge whether to use the historical matching directory according to the similarity; when it is determined to use the historical matching directory, determine the current matching directory according to the historical matching directory and the current keyword group; when it is determined not to use the historical matching directory, determine the current matching directory according to the current keyword group; A processing unit configured to perform cluster analysis on the matching directory and determine a matching display result according to the cluster analysis result; The storage unit is configured to store the current keyword group and the matching display result.

[0006] Furthermore, the acquisition unit performs semantic analysis on the user demand and extracts demand keywords, including: Collect user demand text based on NLP (Natural Language Processing), segment the user demand text, remove stop words and extract stems to obtain valid text; A dependency syntactic analysis model is used to analyze the relationship between words to extract noun phrases, and the noun phrases that conform to the syntactic structure are used as the demand keywords.

[0007] Furthermore, when the acquisition unit determines the importance of the demand keyword according to the number of occurrences of each demand keyword and the keyword coverage, it includes: Classify all historical keywords in the historical demand set to generate several categories, traverse all subject categories, and record whether each demand keyword appears in each subject category. If it appears, it is recorded as 1, otherwise it is recorded as 0; The number of occurrences of the demand keyword in each of the subject categories is added up, and the keyword coverage is obtained by the following formula:

[0008] Wherein, CR represents the coverage of keywords, and takes values ​​of [0, 1]; Km represents the number of times keyword m appears in each of the subject categories, and T represents the number of subject categories.

[0009] Furthermore, the importance of the demand keyword is calculated by the following formula:

[0010] Among them, I(w) represents the importance of demand keyword w, TF(w) represents the number of occurrences of keyword w, DF(w) represents the number of documents in which keyword w appears in the historical demand set, CR(w) represents the coverage of keyword w; α represents the weight coefficient, and its value is [0.1, 0.5].

[0011] Furthermore, the judgment unit compares the keyword group with each historical keyword group in the historical demand set, and judges whether to use the historical matching directory according to the similarity, including: The judgment unit obtains the similarity between the keyword group and each historical keyword group in the historical demand set; When the similarity is greater than the similarity threshold, the judgment unit determines to use the historical matching directory and obtains a similarity difference value according to the similarity and the similarity threshold; When the similarity is less than or equal to a similarity threshold, the judgment unit determines not to use the historical matching directory.

[0012] Furthermore, when the judgment unit determines to use the historical matching directory, it also includes: The judgment unit determines the proportion of the historical matching directory to be used according to the similarity difference; The judgment unit compares the similarity difference with a first preset similarity difference and a second preset similarity difference respectively, and determines the proportion of using the historical matching directory according to the comparison result, wherein the first preset similarity difference is smaller than the second preset similarity difference; When the similarity difference is less than or equal to a first preset similarity difference, the judgment unit determines the ratio to be the first preset ratio; when the similarity difference is greater than the first preset similarity difference and less than or equal to a second preset similarity difference, the judgment unit determines the ratio to be the second preset ratio; when the similarity difference is greater than the second preset similarity difference, the judgment unit determines the ratio to be the third preset ratio; the first preset ratio is less than the second preset ratio, and the second preset ratio is less than the third preset ratio.

[0013] Furthermore, when the judging unit determines not to use the historical matching directory, the following steps are performed: The judging unit screens documents including all the demand keywords in the current keyword group, and collects the number of documents to determine whether to exclude some demand keywords based on the number of documents; When the number of documents is less than the number threshold, the judgment unit removes the required keywords in order from low to high importance until the number of documents is greater than or equal to the number threshold, and uses the screened documents as the current matching directory; When the number of the documents is greater than or equal to the number threshold, the judgment unit determines not to exclude the demand keyword and uses the screened documents as the current matching directory.

[0014] Furthermore, when the processing unit performs cluster analysis on the matching directory, it includes: The processing unit pre-processes each matching document in the matching directory to obtain a vocabulary list; Count the frequency of each word in each matching document and construct a document-word matrix. Each row of the matrix represents a document, each column represents a word, and the matrix elements are the frequency of occurrence of the word. Based on the LDA (Latent Dirichlet Allocation) model, the topic distribution of each matching document and the word distribution of each topic are obtained, including: Randomly assign a topic to each word; Reassign topics to each word according to the following conditional probabilities until the model converges; Get document-topic distribution and topic-word distribution; The conditional probability:

[0015] in, Represents a given document ,word and other words for current topic assignment In the case of The probability of being assigned to topic k; represents the number of words assigned to topic k in document d; represents the Dirichlet prior parameter of the document-topic distribution, represents the total number of all words in document d; K represents the number of topics; represents the number of times word w appears in topic k; Dirichlet prior parameters representing the topic-word distribution; represents the total number of all words in topic k; V represents the total number of words in the vocabulary list.

[0016] Furthermore, when the processing unit performs cluster analysis, it also includes: S1: taking the topic distribution of each matching document as the feature vector of the matching document; S2: Initialize K centroids in all the feature vectors, assign the feature vectors to the nearest centroids, and form K clusters; S3: recalculate the centroid of each cluster; S4: Repeat S2 and S3 until the centroid no longer changes or the preset number of repetitions is completed to obtain the clustering result.

[0017] Furthermore, when the processing unit determines the matching display result according to the cluster analysis result, it includes: When the matching documents after clustering belong to the same category, the matching documents in the current matching directory are sorted and displayed according to the application date from recent to oldest; When the matching documents belong to different categories after clustering, they are displayed in order from high to low according to the number of documents contained in each category and from recent to ancient according to the application date.

[0018] Compared with the prior art, the beneficial effects of the present invention are: by collecting user needs and historical demand sets and performing semantic analysis, extracting demand keywords, and determining their importance based on the frequency of occurrence and coverage of keywords, a more accurate demand expression is achieved. Through the intelligent judgment module, the current keyword group is compared with each historical keyword group in the historical demand set, and whether to reuse the historical matching directory is determined based on the similarity, so as to efficiently match the directory that is highly relevant to the user's needs. The processing unit performs cluster analysis on the matching directory to ensure that the display results are more intuitive and help users quickly obtain the required information. In addition, the storage unit stores the current keyword group and matching results for future calls and optimization. The use of big data analysis technology improves the matching efficiency and accuracy, and solves the problems of low efficiency and low accuracy in traditional intellectual property matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 A functional block diagram of the big data-based intellectual property information analysis and matching system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0021] In some embodiments of the present application, see Figure 1 As shown, a big data-based intellectual property information analysis and matching system includes: a collection unit, a judgment unit, a processing unit and a storage unit. The collection unit is configured to collect user needs and historical demand sets, perform semantic analysis on user needs, extract demand keywords, determine the importance of demand keywords based on the number of occurrences of each demand keyword and the keyword coverage, and sort all demand keywords according to their importance to form a current keyword group.

[0022] The judgment unit is configured to compare the keyword group with each historical keyword group in the historical demand set, and judge whether to use the historical matching directory according to the similarity. When it is determined to use the historical matching directory, the current matching directory is determined according to the historical matching directory and the current keyword group. When it is determined not to use the historical matching directory, the current matching directory is determined according to the current keyword group.

[0023] The processing unit is configured to perform cluster analysis on the matching directory and determine the matching display result according to the cluster analysis result.

[0024] The storage unit is configured to store the current keyword group and the matching display result.

[0025] Specifically, the collection unit collects the demand information input by the user and the historical demand set stored in the system, and performs in-depth semantic analysis on the user's demand. Specifically, the collection unit performs word segmentation and grammatical analysis on the demand input by the user, and extracts the keywords in the demand through natural language processing (NLP) technology. In order to identify the importance of these keywords, the number of keyword occurrences and the coverage of keywords (i.e., the breadth of the subject area affected or involved by the keywords) are used to measure the importance of each keyword. By calculating the importance of these keywords, the keywords are sorted by weight to form a current keyword group. This keyword group is more accurate in expressing the demand and is convenient for subsequent matching analysis. The judgment unit is responsible for comparing the current keyword group with each historical keyword group in the historical demand set to determine their similarity. The text similarity calculation method (such as cosine similarity or Jaccard similarity) is used to calculate the similarity between the current keyword group and each historical keyword group. According to the set similarity threshold, the judgment unit can decide whether to reuse the historical matching directory: if the similarity exceeds the threshold, it is determined that the historical matching directory can be used, thereby avoiding the process of repeatedly analyzing similar requirements. If the similarity is lower than the threshold, the system will generate a new matching directory based on the current keyword group. This judgment mechanism not only improves the matching efficiency, but also reduces the waste of resources caused by repeated calculations. The processing unit performs further cluster analysis on the obtained matching directory. Patent information or intellectual property items with similar requirements are aggregated together through a clustering algorithm to form an easy-to-understand matching display result. This clustering process enables users to quickly find patent information that matches their needs in a more intuitive display interface, avoiding the defects of complex information and difficult selection for users in traditional list displays. The storage unit is responsible for persistent storage of the current keyword group and matching display results. These keyword groups and clustered matching results are saved as structured data to facilitate subsequent calls and reuse of historical data. The system performance is optimized, and historical data can also be used in future demand matching, greatly reducing analysis time.

[0026] It can be understood that through semantic analysis and importance ranking, the accurate expression of user requirements is achieved. The intelligent reuse of historical matching catalogs is realized by using a similarity judgment unit. The display of matching results is optimized through clustering analysis. Finally, historical matching data is saved in the storage unit to support continuous optimization. A closed-loop data flow is formed from requirement collection to matching display, improving the efficiency and accuracy of intellectual property information analysis and matching, and solving problems such as the traditional system relying on subjective judgment, low efficiency, and difficulty in reusing historical data.

[0027] In some embodiments of the present application, when the acquisition unit performs semantic analysis on user requirements and extracts requirement keywords, it includes: Collect user requirement text based on NLP, perform word segmentation, stop word removal, and stemming on the user requirement text to obtain effective text.

[0028] Use a dependency syntax analysis model to parse the relationships between words and extract noun phrases, and use the noun phrases that conform to the syntactic structure as requirement keywords.

[0029] Specifically, for the user requirement text, word segmentation is performed, that is, the continuous text is split into smaller word groups or word units. This step is particularly important when processing Chinese or other languages without obvious word separators, because word segmentation can break a paragraph into words with specific meanings, facilitating subsequent analysis. After word segmentation, the text is filtered for stop words, removing words that frequently appear in the requirements but do not affect semantic understanding (such as "de", "he", etc.). Removing stop words can help retain the core words that are truly information - valuable in the text and reduce noise interference. Stemming is performed on the remaining words, merging different forms with the same root (such as nouns, adjectives, etc.) into a unified stem. For example, both "application" and "applied" are merged into "application", thus effectively reducing word redundancy and improving the consistency of text processing. After obtaining the effective text, a dependency syntax analysis model is further used to identify the grammatical relationships between words in the text. Dependency syntax analysis can judge the dependency relationships between each word and other words, help understand the sentence structure, and extract noun phrases that conform to grammatical rules. Specifically: by parsing the relationships between words through the syntax analysis model, lexical combinations with a master - slave relationship are combined into grammatical noun phrases. These noun phrases usually contain the core elements of user requirements and can accurately express the main intention of the requirements, so they are selected as requirement keywords. For example, in the requirement "A solution to improve data processing speed", noun phrases such as "data processing speed" and "solution" are extracted through syntax analysis as key requirement words to ensure the accurate transmission of requirements.

[0030] It is understandable that through NLP preprocessing and dependency syntactic analysis, not only the keywords representing user needs are accurately extracted, but also a deep understanding of the text structure is achieved, effectively retaining the core semantics of the needs. Compared with simple keyword matching, it can more comprehensively analyze the semantic association of words, especially when there are polysemous words or complex expressions, which improves the accuracy of subsequent demand matching and the intelligent level of system response.

[0031] In some embodiments of the present application, when the collection unit determines the importance of the demand keyword according to the number of occurrences of each demand keyword and the keyword coverage, it includes: Classify all historical keywords in the historical demand set to generate several categories, traverse all subject categories, and record whether each demand keyword appears in each subject category. If it appears, it is recorded as 1, otherwise it is recorded as 0.

[0032] Add up the number of times the demand keyword appears in each subject category and use the following formula to get the keyword coverage.

[0033] ;

[0034] Among them, CR represents the coverage of keywords and takes values ​​of [0, 1]. Km represents the number of times keyword m appears in each topic category, and T represents the number of topic categories.

[0035] In some embodiments of the present application, the importance of the required keywords is calculated by the following formula:

[0036] Among them, I(w) represents the importance of demand keyword w, TF(w) represents the number of occurrences of keyword w, DF(w) represents the number of documents in which keyword w appears in the historical demand set, and CR(w) represents the coverage of keyword w. α represents the weight coefficient, and its value is [0.1, 0.5].

[0037] Specifically, first classify all keywords in the historical demand set and generate several categories (i.e., subject categories) to ensure that each category can represent a specific type of intellectual property demand subject. For each demand keyword, traverse all subject categories and record whether the keyword appears in each category. If the keyword appears in the category, it is marked as 1, otherwise it is marked as 0. By superimposing the occurrence mark values ​​of the keyword in each subject category, the coverage of the keyword is calculated. It is understandable that the accuracy and reliability of demand keyword extraction are effectively improved through comprehensive analysis of the frequency of occurrence, coverage, and document frequency of keywords. Compared with the traditional single word frequency analysis, the weight calculation method of this embodiment is more hierarchical and sophisticated, and can more accurately evaluate the representativeness of keywords in user demand expression. Through this multi-dimensional keyword importance calculation, the accuracy and intelligence level of demand matching are effectively improved, making the intellectual property information matching results more in line with the real needs of users.

[0038] In some embodiments of the present application, the judgment unit compares the keyword group with each historical keyword group in the historical demand set, and determines whether to use the historical matching directory based on the similarity, including: the judgment unit obtains the similarity between the keyword group and each historical keyword group in the historical demand set.

[0039] Specifically, when the similarity is greater than the similarity threshold, the judgment unit determines to use the historical matching directory and obtains the similarity difference according to the similarity and the similarity threshold. When the similarity is less than or equal to the similarity threshold, the judgment unit determines not to use the historical matching directory.

[0040] In some embodiments of the present application, when the judgment unit determines to use the historical matching directory, the method further includes: the judgment unit determines the proportion of the historical matching directory to be used according to the similarity difference. The judgment unit compares the similarity difference with the first preset similarity difference and the second preset similarity difference, respectively, and determines the proportion of the historical matching directory to be used according to the comparison result, and the first preset similarity difference is less than the second preset similarity difference.

[0041] Specifically, when the similarity difference is less than or equal to the first preset similarity difference, the judgment unit determines the ratio to be the first preset ratio. When the similarity difference is greater than the first preset similarity difference and less than or equal to the second preset similarity difference, the judgment unit determines the ratio to be the second preset ratio. When the similarity difference is greater than the second preset similarity difference, the judgment unit determines the ratio to be the third preset ratio. The first preset ratio is less than the second preset ratio, and the second preset ratio is less than the third preset ratio.

[0042] In some embodiments of the present application, when the judgment unit determines not to use the historical matching directory, it includes: the judgment unit screens documents including all the required keywords in the current keyword group, collects the number of documents, and determines whether to eliminate some required keywords based on the number of documents.

[0043] When the number of documents is less than the quantity threshold, the judgment unit removes the required keywords in order of importance from low to high until the number of documents is greater than or equal to the quantity threshold, and uses the filtered documents as the current matching directory.

[0044] When the number of documents is greater than or equal to the number threshold, the judgment unit determines not to exclude the required keyword and uses the screened documents as the current matching directory.

[0045] Specifically, the judgment unit obtains the similarity between the current keyword group and each historical keyword group in the historical demand set. The specific implementation method may calculate the similarity score by methods such as cosine similarity and Jaccard similarity. When the similarity is greater than the set similarity threshold, it is determined that the historical matching directory can be used, and the similarity difference between the two (i.e., the result of subtracting the threshold from the similarity) is recorded. If the similarity is less than or equal to the threshold, it is considered that it is not suitable to use the historical matching directory. When it is determined that the historical matching directory can be used, the use ratio is further determined according to the similarity difference. Through this hierarchical use ratio setting, the degree of use of the historical matching directory can be flexibly adjusted to improve the accuracy and dynamic adaptability of the matching. When it is determined that the historical matching directory is not used, the current keyword group will be screened. Documents containing all the demand keywords in the current keyword group are screened, and the number of documents that meet the conditions is calculated. If the number of documents is less than the set number threshold, the demand keywords are eliminated in order from low to high according to the importance of the keywords until the number of documents reaches or exceeds the threshold. This mechanism ensures that the generated matching directory not only meets the coverage of the current demand, but also avoids overly sparse matching. If the number of documents is greater than or equal to the threshold, there is no need to remove keywords, and the initially screened documents are directly used as the current matching directory.

[0046] It is understandable that the similarity threshold and multi-level ratio access mechanism effectively enhance the intelligence level of intellectual property information matching. By screening layer by layer and flexibly adjusting the construction method of the matching directory, it improves the adaptability to the required keywords, ensuring that the matching results meet both accuracy and adaptability to the diversity of needs. When encountering low similarity or a small number of documents, the elimination strategy can also make the matching directory have sufficient document support, ensuring the stability and reliability of demand matching.

[0047] In some embodiments of the present application, when the processing unit performs cluster analysis on the matching directory, it includes: The processing unit pre-processes each matching document in the matching directory to obtain a vocabulary list.

[0048] Count the frequency of each word in each matching document and construct a document-word matrix. Each row of the matrix represents a document, each column represents a word, and the matrix elements are the frequency of occurrence of the word.

[0049] Based on the LDA model, we obtain the topic distribution of each matching document and the word distribution of each topic, including: Randomly assign a topic to each word.

[0050] Reassign topics to each word based on the following conditional probabilities until the model converges.

[0051] Get document-topic distribution and topic-word distribution.

[0052] Conditional probability: ;

[0053] in, Represents a given document ,word and other words for current topic assignment In the case of The probability of being assigned to topic k. represents the number of words assigned to topic k in document d. represents the Dirichlet prior parameter of the document-topic distribution, represents the total number of words in document d. K represents the number of topics. Indicates the number of times word w appears in topic k. Dirichlet prior parameters representing the topic-word distribution. Represents the total number of all words in topic k. V represents the total number of words in the vocabulary list.

[0054] In some embodiments of the present application, when the processing unit performs cluster analysis, it also includes: S1: Take the topic distribution of each matching document as the feature vector of the matching document.

[0055] S2: Initialize K centroids in all feature vectors and assign feature vectors to the nearest centroids to form K clusters.

[0056] S3: Recalculate the centroid of each cluster.

[0057] S4: Repeat S2 and S3 until the centroid no longer changes or the preset number of repetitions is completed to obtain the clustering result.

[0058] Specifically, the LDA model is: ;

[0059] in, represents the probability of word w appearing in the matching document d. K represents the number of preset topics, represents the probability of word w appearing given topic z, represents the probability of topic z appearing in matching document d.

[0060] Specifically, each matching document in the matching directory is preprocessed to extract the vocabulary list of each document. On this basis, the frequency of each word in each document is counted to construct a document-word matrix. Each row of the matrix represents a document, each column represents a word, and the matrix elements are the frequency of occurrence of the word. This matrix provides a word frequency basis for subsequent LDA topic model analysis. The LDA model is used to extract the topic distribution of the document and the word distribution of each topic. The LDA model randomly assigns an initial topic to each word in the document and iteratively updates it according to the conditional probability formula. Specifically including: Randomly assign a topic to each word. Based on the conditional probability formula, reassign the topic of each word until the model converges. After iterative conditional probability calculation, obtain the document-topic distribution (the probability distribution of each document in different topics) and topic-word distribution (the probability distribution of words under each topic), and establish a topic distribution feature vector for each document.

[0061] It is understandable that by combining the LDA topic model and the K-means clustering algorithm, a document clustering method based on topic distribution is provided. Compared with the clustering method that only uses word frequency or keywords, it achieves topic-level matching, can deeply capture the potential semantic structure of the demand document, and thus more accurately display the demand matching results. It effectively improves the semantic accuracy of the matching and is suitable for the matching analysis of large amounts of intellectual property information.

[0062] In some embodiments of the present application, when the processing unit determines the matching display result according to the cluster analysis result, it includes: When the matching documents belong to the same category after clustering, the matching documents in the current matching directory are sorted and displayed from recent to oldest application dates.

[0063] When the matching documents belong to different categories after clustering, they are displayed in order from high to low according to the number of documents contained in each category and from recent to ancient according to the application date.

[0064] It is understandable that when the clustering results show that all matching documents belong to the same class, it means that these documents have high similarity in content, subject or demand characteristics. In this case, the processing unit sorts these matching documents according to the application date and displays them in order from the latest to the oldest. In this way, users can see the latest matching information first. When the results of the clustering analysis show that the matching documents belong to different categories, they are prioritized for sorting and display according to the number of documents contained in the category, that is, the category containing more documents is prioritized. For the documents within each category, they continue to be arranged in order from the latest to the oldest according to the application date. This method enables the display results to highlight the main categories and reflect the chronological order within each category, ensuring the structure and readability of the display results. By combining the sorting method of the two dimensions of category and time, the orderly display of the clustering analysis results is achieved, so that users can more intuitively obtain the distribution and time priority of matching documents. The readability and user experience of the matching display are effectively improved, especially in a multi-category matching directory, users can quickly focus on key categories and obtain relevant information in chronological order.

[0065] In the above embodiment, by collecting user needs and historical demand sets and performing semantic analysis, demand keywords are extracted, and their importance is determined based on the frequency of occurrence and coverage of the keywords, so as to achieve more accurate demand expression. Through the intelligent judgment module, the current keyword group is compared with each historical keyword group in the historical demand set, and whether to reuse the historical matching directory is determined based on the similarity, so as to efficiently match the directory that is highly relevant to the user's needs. The processing unit performs cluster analysis on the matching directory to ensure that the display results are more intuitive and help users quickly obtain the required information. In addition, the storage unit stores the current keyword group and matching results for future calls and optimization. The use of big data analysis technology improves the matching efficiency and accuracy, and solves the problems of low efficiency and low accuracy in traditional intellectual property matching.

[0066] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0067] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0068] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A big data-based intellectual property information analysis and matching system, characterized in that: include: A collection unit is configured to collect user needs and historical demand sets, perform semantic analysis on the user needs, extract demand keywords, determine the importance of the demand keywords according to the number of occurrences of each demand keyword and the keyword coverage; sort all the demand keywords according to the importance to form a current keyword group; A judgment unit is configured to compare the keyword group with each historical keyword group in the historical demand set, and judge whether to use the historical matching directory according to the similarity; when it is determined to use the historical matching directory, determine the current matching directory according to the historical matching directory and the current keyword group; When it is determined that the historical matching directory is not to be used, determining the current matching directory according to the current keyword group; A processing unit configured to perform cluster analysis on the matching directory and determine a matching display result according to the cluster analysis result; The storage unit is configured to store the current keyword group and the matching display result.

2. The intellectual property information analysis and matching system based on big data according to claim 1 is characterized in that: The collecting unit performs semantic analysis on the user demand and extracts demand keywords, including: Collect user demand text based on NLP, perform word segmentation, remove stop words and extract stems on the user demand text to obtain valid text; A dependency syntactic analysis model is used to analyze the relationship between words to extract noun phrases, and the noun phrases that conform to the syntactic structure are used as the demand keywords.

3. The intellectual property information analysis and matching system based on big data according to claim 1 is characterized in that: When the collecting unit determines the importance of the demand keyword according to the number of occurrences of each demand keyword and the keyword coverage, it includes: Classify all historical keywords in the historical demand set to generate several categories, traverse all subject categories, and record whether each demand keyword appears in each subject category. If it appears, it is recorded as 1, otherwise it is recorded as 0; The number of occurrences of the demand keyword in each of the subject categories is added up, and the keyword coverage is obtained by the following formula: ； Wherein, CR represents the coverage of keywords, and takes values ​​of [0, 1]; Km represents the number of times keyword m appears in each of the subject categories, and T represents the number of subject categories.

4. The intellectual property information analysis and matching system based on big data according to claim 3 is characterized in that: The importance of the demand keywords is calculated by the following formula: ； Among them, I(w) represents the importance of demand keyword w, TF(w) represents the number of occurrences of keyword w, DF(w) represents the number of documents in which keyword w appears in the historical demand set, CR(w) represents the coverage of keyword w; α represents the weight coefficient, and its value is [0.1, 0.5].

5. The intellectual property information analysis and matching system based on big data according to claim 1 is characterized in that: The judgment unit compares the keyword group with each historical keyword group in the historical demand set, and judges whether to use the historical matching directory according to the similarity, including: The judgment unit obtains the similarity between the keyword group and each historical keyword group in the historical demand set; When the similarity is greater than the similarity threshold, the judgment unit determines to use the historical matching directory and obtains a similarity difference value according to the similarity and the similarity threshold; When the similarity is less than or equal to a similarity threshold, the judgment unit determines not to use the historical matching directory.

6. The intellectual property information analysis and matching system based on big data according to claim 5 is characterized in that: When the judging unit determines to use the historical matching directory, it also includes: The judgment unit determines the proportion of the historical matching directory to be used according to the similarity difference; The judgment unit compares the similarity difference with a first preset similarity difference and a second preset similarity difference respectively, and determines the proportion of using the historical matching directory according to the comparison result, wherein the first preset similarity difference is smaller than the second preset similarity difference; When the similarity difference is less than or equal to a first preset similarity difference, the judgment unit determines the ratio to be the first preset ratio; when the similarity difference is greater than the first preset similarity difference and less than or equal to a second preset similarity difference, the judgment unit determines the ratio to be the second preset ratio; when the similarity difference is greater than the second preset similarity difference, the judgment unit determines the ratio to be the third preset ratio; the first preset ratio is less than the second preset ratio, and the second preset ratio is less than the third preset ratio.

7. The intellectual property information analysis and matching system based on big data according to claim 5 is characterized in that: When the judging unit determines not to use the historical matching directory, it includes: The judging unit screens documents including all the demand keywords in the current keyword group, and collects the number of documents to determine whether to exclude some demand keywords based on the number of documents; When the number of documents is less than the number threshold, the judgment unit removes the required keywords in order from low to high importance until the number of documents is greater than or equal to the number threshold, and uses the screened documents as the current matching directory; When the number of the documents is greater than or equal to the number threshold, the judgment unit determines not to exclude the demand keyword and uses the screened documents as the current matching directory.

8. The intellectual property information analysis and matching system based on big data according to claim 1, characterized in that: When the processing unit performs cluster analysis on the matching directory, it includes: The processing unit pre-processes each matching document in the matching directory to obtain a vocabulary list; Count the frequency of each word in each matching document and construct a document-word matrix. Each row of the matrix represents a document, each column represents a word, and the matrix elements are the frequency of occurrence of the word. Based on the LDA model, we obtain the topic distribution of each matching document and the word distribution of each topic, including: Randomly assign a topic to each word; Reassign topics to each word according to the following conditional probabilities until the model converges; Get document-topic distribution and topic-word distribution; The conditional probability: ； in, Represents a given document ,word and other words for current topic assignment In the case of The probability of being assigned to topic k; represents the number of words assigned to topic k in document d; represents the Dirichlet prior parameter of the document-topic distribution, represents the total number of all words in document d; K represents the number of topics; represents the number of times word w appears in topic k; Dirichlet prior parameters representing the topic-word distribution; represents the total number of all words in topic k; V represents the total number of words in the vocabulary list.

9. The intellectual property information analysis and matching system based on big data according to claim 8, characterized in that: When the processing unit performs cluster analysis, it also includes: S1: taking the topic distribution of each matching document as the feature vector of the matching document; S2: Initialize K centroids in all the feature vectors, assign the feature vectors to the nearest centroids, and form K clusters; S3: recalculate the centroid of each cluster; S4: Repeat S2 and S3 until the centroid no longer changes or the preset number of repetitions is completed to obtain the clustering result.

10. The intellectual property information analysis and matching system based on big data according to claim 9, characterized in that: When the processing unit determines the matching display result according to the cluster analysis result, it includes: When the matching documents after clustering belong to the same category, the matching documents in the current matching directory are sorted and displayed according to the application date from recent to oldest; When the matching documents belong to different categories after clustering, they are displayed in order from high to low according to the number of documents contained in each category and from recent to ancient according to the application date.

Images