A method and system for associating Chinese and foreign company names

Abstract

This application discloses a method and system for tagging the association of Chinese and foreign company names. The method first preprocesses the raw data, then uses the Spacy NLP tool to perform part-of-speech analysis and tagging on the preprocessed data, identifying the nature of keywords and assigning corresponding tags. The data is then re-tagged based on a preset surname filter list. The re-tagged results are integrated and inaccurate tags are removed. The inverse document frequency values ​​of the keywords are then calculated to determine brand terms. Finally, the final association between Chinese and foreign company names is determined based on the brand terms and the results are output. By combining natural language processing technology with preset rules, the present invention achieves efficient and accurate company name association tagging without the need for an external industrial and commercial database, significantly reducing costs and labor input.

Description

Technical Field

[0001] The present application relates to the technical field of text extraction and processing, specifically to the fields of natural language processing and deep learning, and in particular to a method and system for associating Chinese and foreign company names. Background Art

[0002] In the patent field, identifying and extracting corporate brand terms and mapping them to other languages ​​has long been a technical challenge for major patent application companies. As companies expand their operations across different regions, their names have evolved into various language variations, such as "Group," "Stock Co., Ltd.", "LTD," "AB," and "gmbh."

[0003] Traditional methods typically involve purchasing expensive databases or manually annotating these company names to process and link them. However, these methods are not only costly but also time-consuming and labor-intensive. Furthermore, the complexity of patent priority data further complicates the process of linking Chinese and English company names, as this data may include collaborations between companies and affiliates, individuals, and parent companies and subsidiaries. Therefore, an efficient, accurate, and cost-effective method for linking Chinese and foreign company names is urgently needed to meet the needs of aggregated company queries in the patent field. Summary of the Invention

[0004] This application provides a method and system for associating Chinese and foreign company names. Through data preprocessing, natural language processing, name recognition and other processes, it outputs relatively accurate mapping relationships between Chinese and English, and Chinese and small languages ​​worldwide, establishes a standard knowledge base, and facilitates aggregated queries on corporate patents.

[0005] In a first aspect, a method for associating Chinese and foreign company names is provided, the method comprising:

[0006] Preprocess the raw data according to the preset order and normalization method;

[0007] Use Spacy NLP tools to perform part-of-speech judgment and tagging on the pre-processed data, identify the nature of keywords and mark them accordingly;

[0008] The data is secondary marked based on the preset surname filter list, and the name matching rule is judged to form the secondary marking result;

[0009] Integrate the secondary tagging results and remove inaccurate tags; wherein the inaccurate tags include removing preset common words;

[0010] Calculate the inverse document frequency value of keywords and determine brand words;

[0011] Determine the final association between Chinese and foreign company names based on brand words and output the results.

[0012] Optionally, the raw data is preprocessed according to a preset order and normalization method, specifically including:

[0013] Convert lowercase letters in English names to uppercase letters;

[0014] Clean up the spaces at the beginning and end of the name;

[0015] Replace more than one space with a single space;

[0016] Remove spaces between Chinese characters;

[0017] Remove special characters.

[0018] Optionally, use the Spacy NLP tool to perform part-of-speech detection and tagging on the preprocessed data, including:

[0019] Call Spacy's nlp method to perform part-of-speech judgment on the preprocessed English keywords, identify the nature of the keywords, and mark them accordingly; the nature of the keywords includes company, organization, or person name.

[0020] Optionally, the data is re-labeled based on a pre-set surname filter list, including:

[0021] Collect surnames commonly used in major regions and add them to the filter list;

[0022] The first character, the last character, and the first 2, 3, and 4 characters of the keyword are intercepted and added to the judgment list;

[0023] Compare each character with the surname list one by one. If at least one character in the list matches the surname list, the keyword is considered to be a surname and the tag "PER" is returned.

[0024] Optionally, the secondary labeling results are integrated, including:

[0025] The labeling results of the Spacy NLP tool and the secondary labeling results were cross-validated. When the labeling results of the two steps were completely consistent and both were "PER", the label was considered accurate. The data marked as human names were eliminated.

[0026] Optionally, the step of calculating the inverse document frequency value of the keyword includes:

[0027] Organize and eliminate commonly used meaningless words;

[0028] according to Calculate the IDF value of each English keyword, where N is the total number of documents in the document set and df(t) is the number of documents containing word t;

[0029] Output the keyword with the smallest IDF value as the brand word.

[0030] In the second aspect, a system for associating Chinese and foreign company names is provided, the system comprising:

[0031] A preprocessing module is used to preprocess the raw data according to a preset order and normalization method;

[0032] The formal processing module is used to use the Spacy NLP tool to perform part-of-speech judgment and tagging on the pre-processed data, identify the nature of keywords and mark them accordingly; perform secondary tagging on the data based on a preset surname filter list, perform name matching rule judgment to form secondary tagging results; integrate the secondary tagging results and eliminate inaccurate tags; wherein, the inaccurate tags include eliminating preset common words;

[0033] The storage module is used to calculate the inverse document frequency value of keywords and determine brand words;

[0034] The enterprise name query module is used to determine the final association relationship between Chinese and foreign enterprise names based on brand words and output the results.

[0035] In a third aspect, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method for associating Chinese and foreign company names as described in any one of the first aspects above is implemented.

[0036] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for associating Chinese and foreign company names as described in any one of the first aspects is implemented.

[0037] In a fifth aspect, a computer program product is provided, comprising a computer program / instruction, which, when executed by a processor, implements the method for associating Chinese and foreign company names as described in any one of the first aspects.

[0038] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least:

[0039] (1) By utilizing natural language processing tools and a pre-set filter list, we achieved automatic association tagging of Chinese and foreign company names. Compared with traditional manual tagging methods, this method does not require extensive human intervention, significantly reducing the workload and time investment of manual tagging. At the same time, it avoids the subjective errors and inconsistencies that may occur in the manual tagging process, and improves the efficiency and accuracy of company name association.

[0040] (2) Through a multi-step processing process, including pre-processing, part-of-speech determination, secondary tagging, integration of tag results, and calculation of inverse document frequency (IDF) values, the data is screened and verified layer by layer. This comprehensive processing method can effectively eliminate personal names, common words, and meaningless words, ensuring the accuracy of the association between brand words and company names in the final output. After testing, the accuracy of this method can be maintained at above 95%, meeting the daily needs of company name alignment and tagging in the patent field. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other implementation drawings based on the provided drawings without inventive effort.

[0042] Figure 1 A flowchart of the steps of the method for associating Chinese and foreign company names provided in the embodiment of this application;

[0043] Figure 2 A flowchart provided for an optional embodiment of the present application;

[0044] Figure 3 A schematic diagram of the system architecture provided in an embodiment of the present application;

[0045] Figure 4 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0046] In the description of the present application: the terms "comprise", "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units that have been explicitly listed, but may also include other steps or units that are not explicitly listed but are inherent to these processes, methods, products or apparatuses, or steps or units that are added based on further optimization solutions conceived by the present invention.

[0047] The identification, extraction, and mapping of corporate brand terms in other languages ​​within the patent field have long been a technical challenge for major patent application vendors. As businesses expand across different regions, they generate a variety of corporate names and variants in different languages, such as Group, Stock Corporation, LTD, AB, and GmbH. Conventional companies typically rely on expensive World Business Relations Databases or manual annotation to process and map these names, requiring significant time and cost.

[0048] Judging from the priority data, the data composition is complex. Some companies and affiliated companies jointly wrote the data, some individuals wrote the data independently, and some parent companies and subsidiaries jointly wrote the data. This makes it extremely difficult to associate Chinese and multilingual company names.

[0049] like Figure 1 As shown, a method for associating Chinese and foreign company names is provided, which specifically includes:

[0050] S1, preprocessing the original data according to the preset order and normalization method.

[0051] In this step, the raw data is first preprocessed to ensure data consistency and processability. The specific process includes converting lowercase letters in English names to uppercase letters to eliminate the impact of case differences on subsequent processing; cleaning up spaces at the beginning and end of names to avoid data inconsistencies caused by excess spaces; replacing more than one consecutive space with a single space to further standardize the data format; removing spaces between Chinese characters to adapt to the processing requirements of Chinese text; and removing special characters such as quotation marks, periods, commas, dashes, etc. These special characters may interfere with subsequent natural language processing steps. Through these normalization operations, the raw data is converted to a unified format, laying the foundation for subsequent processing steps.

[0052] S2, uses the Spacy NLP tool to perform part-of-speech judgment and tagging on the preprocessed data, identify the nature of the keywords and mark them accordingly.

[0053] Based on the preprocessed data, the present invention uses the Spacy NLP tool to perform part-of-speech (POS) identification and tagging. Spacy is a widely used natural language processing (NLP) library that effectively identifies keywords and their properties in text. In the specific implementation, Spacy's NLP method is called to analyze the preprocessed English keywords, identifying their parts of speech (POS), such as "company," "organization," and "person," and assigning a corresponding tag to each keyword. These tags serve as an important basis for determining whether a keyword is a company name in subsequent steps.

[0054] S3, based on the preset surname filter list, the data is secondary marked, and the name matching rule is judged to form the secondary marking result.

[0055] The present invention introduces a secondary tagging step based on a surname filtering list. In this step, commonly used surnames in major regions are first collected to form a preset surname filtering list. The preprocessed data is then subjected to name matching rule judgment. Specifically, the first character, last character, and first two, three, and four characters of the keyword are intercepted to form a judgment list. These intercepted characters are then compared one by one with the surname list. If at least one character in the judgment list matches the surname list, the keyword is considered likely to be a name, and the tag "PER" is returned.

[0056] S4, integrate the secondary labeling results and remove inaccurate labels.

[0057] Among them, the inaccurate marking includes eliminating preset common words. After completing the part-of-speech judgment based on the Spacy NLP tool and the secondary marking based on the surname filtering list, the present invention integrates the two marking results. The specific process is to cross-validate the marking results of the Spacy NLP tool with the secondary marking results. When the marking results of the two steps are completely consistent and are both "PER", the label is considered to be accurate; otherwise, it is regarded as an inaccurate mark and is eliminated. In addition, this step also includes eliminating preset common words, which are usually words that have no practical meaning, such as "CORP", "THE", "INC", "UNIV", "CO", "LTD", "CHINA", "OF", "TECHNOLOGY", etc.

[0058] S5, calculate the inverse document frequency value of the keyword and determine the brand word.

[0059] After integrating the tagging results and removing common words, the present invention determines brand words by calculating the inverse document frequency (IDF) value of the keywords. In the specific implementation process, common and meaningless words are first sorted and removed to ensure that the dataset contains only meaningful keywords.

[0060] Then, according to Calculate the IDF value of each English keyword, where N is the total number of documents in the document set and df(t) is the number of documents containing word t. Finally, select the keyword with the smallest IDF value as the brand word because these keywords have a high degree of discrimination in the dataset and are more likely to be the core part of the company name.

[0061] S6, determine the final association relationship between Chinese and foreign company names based on the brand words and output the result.

[0062] The present invention traverses the mapping set to check whether the English keyword contains the brand word determined in step S5. If an English keyword contains the brand word, it is considered that the English keyword has established a matching relationship with the Chinese keyword. Ultimately, these matching relationships are output as the association results of the Chinese and foreign company names.

[0063] like Figure 2 , a flowchart of an optional implementation is given: the implementation involves the following steps or modules:

[0064] Step S1, pre-processing module: pre-process the original data to be processed according to a specific order and normalization method, pre-process the priority data according to the "Chinese-English" format, and use the mapping set A{CnKey:List <string>L} after the following processing, it becomes: A'{CnKey:List <string>L'}, the preprocessing specifically includes:

[0065] Convert lowercase to uppercase;

[0066] Clean up the first and last spaces;

[0067] Brackets are organized into English;

[0068] More than one space is replaced by one space;

[0069] Remove spaces between Chinese characters;

[0070] Special characters are removed.

[0071] Sort the priority data by the original Chinese-English mapping relationship, and arrange the Chinese names in positive order to form one: Chinese name, and multiple English or other language names to facilitate the next step of processing.

[0072] In step S2, based on Spacy NLP (a natural language processing tool commonly used in Python), the properties of English keywords are identified and filtered, such as company, organization, and person's name. The dataset L' is judged for the following parts of speech and marked to form:

[0073] L'':List <tuple<en_name ,property_name1> >

[0074] English keyword part of speech judgment:

[0075] Call the Spacy NLP method to output a limited number of tags, such as company, organization, and person. Based on the identified non-core word tags such as names of people and places, filter and filter non-company and organization entity words. The above tool has good support for English, but has poor support for Chinese and other non-English words, so further processing is required.

[0076] Step S3: Secondary tagging. Collect common surnames from various regions and add them to the filter list. L' performs the following part-of-speech determination and tags, forming L''': Name matching rule: Truncate the first character and add it to the judgment list. Truncate the last character and add it to the judgment list. Truncate the first two, first three, and first four characters and add them to the judgment list. Iterate through the judgment list and compare each character with the surname list. If at least one character in the judgment list matches the surname list, it is considered a surname and the tag "PER" is returned.

[0077] Step S4, integrating label results,

[0078] Cross-validate the results L'' and L''' of S2 and S3. When the labeling results of S2 and S3 are completely consistent and both are "PER", the label is considered accurate, that is:

[0079]

[0080] The data labeled 'PER', that is, the data marked as a person's name, is removed.

[0081] Step S5: remove common words and meaningless words.

[0082] It is necessary to eliminate common words. Here we have sorted out common meaningless words so that after calculating the IDF value, we can get truly meaningful words.

[0083] Take L''''=List<en_name : string>

[0084] Keep other keywords and form A'{CnKey : List <string>L''''}

[0085] Step S6: Calculate the IDF score for the filtered result set.

[0086]

[0087] N is the total number of documents in the document set. df(t) is the number of documents containing the term t.

[0088] After the calculation is completed, it is formed: L5=List<en_name: string , idf_vaue : double > .

[0089] Step S7: Calculate the minimum IDF value and output the keyword corresponding to the minimum value, that is, output the top 1 keyword with the minimum value as the brand word. brand_word=[\min_{(e, i) \in L_5} i].

[0090] Step S8: traverse the one-to-many mapping table, and the one containing the brand word is the final labeling result.

[0091] Traverse A'{CnKey : List <string>L'}, all the English words in brand_word are considered to be the result of establishing a true matching relationship with CnKey.

[0092] In one embodiment, Figure 3 As shown, a system for associating Chinese and foreign company names is provided, the system comprising:

[0093] A preprocessing module is used to preprocess the raw data according to a preset order and normalization method;

[0094] The formal processing module is used to use the Spacy NLP tool to perform part-of-speech judgment and tagging on the pre-processed data, identify the nature of keywords and mark them accordingly; perform secondary tagging on the data based on a preset surname filter list, perform name matching rule judgment to form secondary tagging results; integrate the secondary tagging results and eliminate inaccurate tags; wherein, the inaccurate tags include eliminating preset common words;

[0095] The storage module is used to calculate the inverse document frequency value of keywords and determine brand words;

[0096] The enterprise name query module is used to determine the final association relationship between Chinese and foreign enterprise names based on brand words and output the results.

[0097] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities, and the network interface is used to communicate with an external terminal via a network connection. The computer device loads and runs a computer program to implement the above-mentioned method for associating Chinese and foreign company names.

[0098] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0099] In one embodiment, a computer-readable storage medium is further provided, on which a computer program is stored, which involves all or part of the processes in the above-mentioned embodiment method.

[0100] In one embodiment, a computer program product is also provided, including a computer program / instruction, which involves all or part of the process in the above embodiment method.

[0101] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in M ​​forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (SyMchliMk) DRAM (SLDRAM), memory bus (RaMbus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0102] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.< / string> < / string> < / string> < / string>

Claims

1. A method for associating Chinese and foreign company names, characterized in that: The method comprises: Pre-processing the original data according to a preset order and normalization method; wherein the original data includes enterprise name data; Use Spacy NLP tools to perform part-of-speech judgment and tagging on the pre-processed data, identify the nature of keywords and mark them accordingly; The data is secondary marked based on the preset surname filter list, and the name matching rule is judged to form the secondary marking result; Integrate the secondary tagging results and remove inaccurate tags; wherein the inaccurate tags include removing preset common words; Calculate the inverse document frequency value of the keyword and select the keyword with the smallest inverse document frequency value as the brand word; Determine the final association between Chinese and foreign company names based on the brand word and output the result; wherein, by traversing the mapping set, check whether the English keyword contains the determined brand word. If the English keyword contains the brand word, it is considered that the English keyword and the Chinese keyword have established a matching relationship, and the matching relationship is output as the association result between the Chinese and foreign company names; Among them, the nature of keywords includes company, organization or person names; The data is re-labeled based on a preset surname filter list. This includes: collecting common surnames from major regions and adding them to the filter list; extracting the first, last, first two, third, and fourth characters of the keyword and adding them to the judgment list; and comparing each character against the surname list. If at least one character in the judgment list matches the surname list, the keyword is considered a surname and the label "PER" is returned. The secondary labeling results were integrated, including: cross-validation of the labeling results of the Spacy NLP tool and the secondary labeling results; when the labeling results of the two steps were completely consistent and both were "PER", the label was considered accurate; and data marked as human names were eliminated.

2. The method according to claim 1, characterized in that The raw data is preprocessed according to the preset order and normalization method, including: Convert lowercase letters in English names to uppercase letters; Clean up the spaces at the beginning and end of the name; Replace more than one space with a single space; Remove spaces between Chinese characters; Remove special characters.

3. The method according to claim 1, characterized in that Use Spacy NLP tools to perform part-of-speech judgment and tagging on the preprocessed data, including: Call Spacy's nlp method to perform part-of-speech judgment on the preprocessed English keywords, identify the nature of the keywords, and mark them accordingly.

4. The method according to claim 1, wherein The steps for calculating the inverse document frequency value of a keyword include: Organize and eliminate commonly used meaningless words; according to Calculate the IDF value of each English keyword, where N is the total number of documents in the document set, df ( t ) is a word containing t The number of documents; Output the keyword with the smallest IDF value as the brand word.

5. A system for associating Chinese and foreign company names, characterized in that: The system comprises: A preprocessing module is used to preprocess the original data according to a preset order and normalization method; wherein the original data includes enterprise name data; The formal processing module is used to use the Spacy NLP tool to perform part-of-speech judgment and tagging on the pre-processed data, identify the nature of keywords and mark them accordingly; perform secondary tagging on the data based on a preset surname filter list, perform name matching rule judgment to form secondary tagging results; integrate the secondary tagging results and eliminate inaccurate tags; wherein, the inaccurate tags include eliminating preset common words; A storage module is used to calculate the inverse document frequency value of the keyword and select the keyword with the smallest inverse document frequency value as the brand word; The company name query module is used to determine the final association relationship between Chinese and foreign company names based on brand terms and output the result. It traverses the mapping set to check whether the English keyword contains the determined brand term. If the English keyword contains the brand term, it is considered that the English keyword and the Chinese keyword have established a matching relationship, and the matching relationship is output as the association result of the Chinese and foreign company names. Among them, the nature of keywords includes company, organization or person names; The data is re-labeled based on a preset surname filter list. This includes: collecting common surnames from major regions and adding them to the filter list; extracting the first, last, first two, third, and fourth characters of the keyword and adding them to the judgment list; and comparing each character against the surname list. If at least one character in the judgment list matches the surname list, the keyword is considered a surname and the label "PER" is returned. The secondary labeling results were integrated, including: cross-validation of the labeling results of the Spacy NLP tool and the secondary labeling results; when the labeling results of the two steps were completely consistent and both were "PER", the label was considered accurate; and data marked as human names were eliminated.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images