TW202349325A - A system of semantic analysis-based trademark class recommendation and the method thereof - Google Patents

Abstract

The present invention provides an online application system with a classification recommend module. The system includes an electronic device operated by a user, which connects to a server through internet. The electronic device comprises a processor, a memory, and a network interface controller. The server includes an application program. The user operates the electronic device to execute the application program's trademark online application module, conducting a trademark application online. The user inputs the required data for the trademark application through the electronic device. The system utilizes the classification recommend module and the risk assessment module to generate classification recommendation reports and risk assessment reports, providing solutions to the user's difficulties in selecting classifications.

Description

A semantic analysis trademark category recommendation system and its method

The present invention relates to an online trademark application and management system, especially a system with trademark category recommendation and risk assessment.

In the traditional trademark application process, whether domestic or foreign, it is necessary to print out paper documents and fill in a large number of forms, which is not in line with the principles of environmental protection. Due to the formality of trademark application documents, if the applicant fills them out incorrectly, the entire process will be disrupted. It may need to be done again, which results in a waste of time and human resources, especially when trademark applications are made in other countries. In addition to traditional paperwork, if communication is carried out by personnel, since they have different professional backgrounds, different Language usage, cultural differences, and other unpredictable factors can convey inaccurate or misunderstood information, which results in cognitive differences between applicants and foreign agents and government agencies. As a result, applicants may not obtain the results they originally desired.

On the other hand, in the past, the target users of trademark search and retrieval were professional practitioners. For people without relevant professional background, it was very difficult to understand the search logic. Users could not intuitively understand whether the trademark they wanted to apply for was approved. What are the risks and how to evaluate whether you can safely apply for a trademark or whether you need to make design adjustments to the application text or images. For trademark searches, official search pages established by government agencies of various countries or private trademark search platforms often only list similar previous trademark cases during the trademark search process, and do not provide any result analysis, reference points or related evaluations. , which makes it difficult for users who know nothing about intellectual property rights to judge the impact of these previous cases on their own trademark applications, and it may also be difficult for users to estimate the degree of risk.

In addition, in order to adapt to the different definitions and interpretations of goods and services in various countries, the culture and customs of each country are different. However, the current way to select trademark classification is to search through precise or vague words based on national classification standards and the internationally accepted Nice Classification. to find similar results. However, for users who do not understand the classification standards published by official agencies of various countries or the international Nice Classification description method, it will be very difficult to find correct and corresponding results, and users may miss some The important items they really need. For example, the owner of a clothing store may not know that what he really needs is "wholesale and retail clothing", but just finds some items in the category by himself, such as "women's clothing, men's clothing, shirts" ”. Therefore, in this state, users can only obtain selection suggestions on trademark classification and product items for their trademark applications through traditional manual consultation, but cannot directly apply for trademarks based on their own circumstances.

As a result, merchants cannot easily select trademark categories and file trademark applications in various countries. In the end, users (merchants) can only seek help from local law firms to select appropriate trademark categories and product items in a single country. Seek some professional advice and entrust the implementation of the application case. Generally speaking, it takes one to several weeks from communication to receiving the final category proposal. If it is multiple countries or the proposal is wrong, multiple back and forth often lead to missed opportunities. The traditional process not only increases the complexity of the pre-application process, but also increases the consulting time of a large number of trademark professionals, which in turn leads to high costs for trademark applications, and the pre-application process and execution cycle are also longer. , which in turn affects the general public, when they come up with ideas for a new brand name, slogan or business (product) name, it is easy for them to get lucky and give up on the trademark application due to the difficulty and cost threshold of trademark application, thereby setting the stage for possible subsequent trademark disputes. The incident laid the foundation and irreversible risks.

After the epidemic, it coincides with the revolutionary era of e-commerce and AI development. Cross-border e-commerce transactions and cross-border services are booming. More merchants need cross-border brands. For trademark applicants who have a cross-border layout in trademark applications, When needed, it is usually because the content of trademarked goods items in different countries is different, and it is impossible to directly convert the trademark application content of the original application country into the trademark application content of another country; at this time, it is necessary to entrust professionals again based on the original case. Product items are entered into the target country's product item database for manual comparison and selection. This requires a lot of manpower and back-and-forth confirmation, which not only requires higher execution cycles and costs, but also requires manual In this way, trademark professionals often make mistakes in understanding or recommended operations because they are not experts in the industry, which increases the risk of subsequent corrections and slows down the speed of trademark applications. This results in Increased enforcement costs when applying for a trademark abroad. On the other hand, for trademark practitioners, the repetitive work of trademark recommendation leads to a decrease in job satisfaction and an increase in burnout. Normally, a cross-border trademark recommendation involves communication, understanding, judgment, and the creation of a recommendation letter. The processing can take several hours to several days. Often, after the recommendation letter is completed, it is provided to the trademark applicant, who ultimately fails to apply. Only trademark practitioners can appreciate the amount of energy consumed.

At present, some manufacturers or official systems have developed systems and functions for trademark category recommendation and trademark retrieval. However, they still have various shortcomings in practical use, resulting in users being unable to intuitively Obtain trademark application suggestions and evaluation content based on user experience to facilitate users in the subsequent (multi-country) trademark application process; the following lists and explains the unresolved issues one by one.

First of all, when creators use the LegalZoom platform (https://www.legalzoom.com/business/intellectual-property/trademark-registration-overview.html) provided by LegalZoom.com, Inc. in the United States, LegalZoom’s Although the trademark application service process has been implemented online, more specifically, it is a simple online order-taking process, that is, when issuing a trademark application request, you need to complete the trademark application form provided by the platform, as shown in Figure 1 ~3, including providing information such as the trademark name to be applied for, the scenario in which the trademark name is used, and the user's industry description; after completing the above information filling and payment, the system will send the information to the backend and the specialist assigned by LegalZoom will follow up. Manual offline and consulting services, and then complete the subsequent trademark application services and processes; in the above service process, the LegalZoom platform only realizes and replaces the traditional trademark affairs through business personnel face-to-face or through online orders. The process of conducting offline contact and receiving customer application requirements. However, in the above process, users are still unable to obtain trademark category proposals and implement the trademark application process by themselves. They still need to go through a lengthy interview process.

As disclosed in Chinese Patent Publication No. CN109800340A, please refer to Figure 4 and Figure 5. The user needs to enter the correct text or words in the trademark name field first. The user finally recommends at least one product category based on the product label information. At least one product/service item corresponding to each product category includes: the user finally determines the first similarity between the trademark and trademark information of each trademark category in at least one product label category in the database based on the product label information; and the user Finally, based on the first similarity, at least one product/service item corresponding to each product category in the smallest trademark category is recommended. Determining a second degree of similarity between trademarks and trademark information of similar groups in at least one group, wherein at least one commodity/service corresponding to each commodity category in at least one commodity category is recommended based on the first similarity The items include: recommending at least one product/service item corresponding to each product category in at least one product category based on the first similarity and the second relative degree. For example, the trademark category determined by the user is Category 7. If the first group in Category 7 and the second group in Category 12 are similar groups, the product label that combines the trademark information with the product label in Category 7 is removed. In addition to comparing and determining the first appearance, it is also necessary to further compare the trademark information with the trademarks of the appearance group (i.e. the second group) in the 12th class trademark category and determine the second degree of similarity. Further, goods/services are recommended based on the first degree of similarity and the second degree of similarity. For example, the first group does not have the same trademark or similar trademarks, that is, the first degree of similarity, and the registration success rate is high, but the second group has many similar trademarks and the same trademark, that is, the second degree of similarity is relatively high, and the registration success rate is relatively low. Therefore, considering the first and second similarities, the registration success rate of the first group is relatively low, so it is not recommended to register the first group.

However, the above-mentioned Chinese patent (publication number CN109800340A) has several problems that need to be improved. For example, the trademark recommendation comparison is mainly based on label information. If the label setting is not precise enough or the same product item may match multiple labels at the same time, in the label comparison Therefore, the process may be inaccurate or a large number of unqualified items may be compared, resulting in reduced recommendation benefits. In addition, the above-mentioned Chinese patent uses the group relationship of category items in the Chinese trademark classification as one of the basis for comparison. Although the group relationship in the trademark category is the basis for official search and retrieval, it is not the case for merchants or trademarks. When practitioners need to know the recommended categories, it may also result in items that do not match the comparison, resulting in an excessive amount of data, reducing the effectiveness of recommendations. Ultimately, when the above-mentioned Chinese patent receives user information, the user must input precise and accurate text in the corresponding field, such as a trademark name, so that the tags can be compared and generated to recommend or not recommend registration. result information.

As disclosed in Chinese Patent Announcement No. CN109902196B, a trademark category recommendation method is applied to a trademark category recommendation device. Specifically, the method includes: constructing updated data in real time; using the updated data to compare historical trademark databases The trademark category is updated, wherein the trademark category in the historical trademark database is a historical recommended trademark category; the updated trademark category is determined as a recommended trademark category. In this regard, the trademark category recommendation device constructs updated data in real time, and the trademark category recommendation device updates the trademark categories in the historical trademark database based on the updated data, wherein the trademark categories in the historical trademark database are historical recommendations. Trademark category; then, the updated trademark category is determined as the recommended trademark category, where the recommended trademark category is the trademark category with the smallest registration frequency so far, which can help the user determine the trademark category to be registered, and the user You can also continue to select the trademark category with the smallest registration frequency from the recommended trademark categories.

The Chinese patent with announcement number CN109902196B has problems that need to be improved. The recommendation principle is based on historical recommendation records and statistical registration frequency records of each trademark category. It ensures that the data in the database is up-to-date by continuously updating the data. However, the recommendation system recommends For trademark categories with low frequency or official registration frequency, such a recommendation model is obviously not necessarily the most suitable for users, because even the smallest number of applications will not be of interest to the user. Furthermore, even though the Chinese patent with announcement number CN109902196B will conduct a similarity comparison between the applicant's industry field and the trademark category to find trademark categories in similar fields with low registration rates, in fact it does not provide the applicant with any substantive recommended categories or even categories. It is the benefit of commodity items. Finally, when the Chinese patent with announcement number CN109902196B receives user information, the applicant must enter accurate and accurate text in the corresponding fields, such as trademark name and industry field, so that subsequent industry similarity can be carried out. Compare and identify trademark categories with fewer recommendations and less frequent registrations.

As disclosed in Chinese Patent Publication No. CN111898022A, trademark category recommendation method and device, as well as storage media and electronic equipment, the industry requested by the user is received; the identification information of the industry requested by the user is determined; and the industry based on the industry requested by the user is determined. Identify information, obtain at least one trademark category corresponding to the industry requested by the user, and recommend the at least one trademark category to the user. Optionally, in an embodiment, obtaining at least one trademark category corresponding to the industry requested by the user based on the identification information of the industry requested by the user includes: based on the identification information of the industry requested by the user, The trademark registration category statistical information of the industry requested by the user is searched from the database; where the trademark registration category statistical information of the industry is used to represent the statistical information on the number of each trademark category registered in each industry. Optionally, the statistical information of trademark registration categories in the industry includes: determining the industry to which each company belongs based on the subject information of each company in the database; obtaining all the trademark information based on the trademark information of each company in the database. Based on the industry to which each company belongs and the trademark registration category of each company, statistics are made on the trademark registration categories of each industry to obtain statistical information on the trademark registration categories of each industry. Recommending the at least one trademark category to the user includes: determining at least one trademark registration category for each industry based on statistical information on trademark registration categories for each industry and preset category threshold information for each industry; Based on at least one trademark registration category of each industry, at least one trademark category of the industry requested by the user is recommended to the user.

If the above-mentioned Chinese patent (publication number CN111898022A) still has problems that need to be improved, simply use statistical methods to conduct statistics on the company's service (product) items registered in various public and commercial bureaus in China and the company's trademark application categories, and store them into a database. This model is not applicable to other countries, let alone cross-border category recommendation analysis; when the user selects a specific industry, the statistical results of the trademark application category corresponding to the industry will be jumped out, and a recommendation result will be formed; it is simply recommended for trademark categories , there is no actual product item content recommendation; in fact, the usability is not high. Even if the user barely finds and selects his company's industry from the industry classification provided by the system, he can only get very general "trademark category" information, but In practice, there is a high probability that it is not possible to apply for a single category of all-product items. Therefore, it is recommended but has limited practical effect in assisting customers with trademark applications.

As disclosed in Chinese Patent Publication No. CN107330109A, sample trademark images and content are processed into trademark cards according to preset trademark carding standards. The specific processing process includes: (1) Establishing preset shape features, pronunciation features and A trademark classification standard composed of various combinations of the smallest unit of meaning characteristics, (2) identify whether the sample trademark is composed of Chinese characters, graphics, letters, digits or symbols, and obtain the content of the constituent elements, (3) each of the sample trademarks The smallest unit of shape characteristics, the smallest unit of pronunciation characteristics and the smallest unit of meaning characteristics of the constituent elements; (4) According to the established trademark classification standard, extract the segmentation information of various texts and graphics generated or converted by each combination scheme, Use this segmentation information as sample trademark classification information, and set the similarity evaluation score for each preset trademark classification standard. Use the input trademark card information set as the search keyword to retrieve the sample trademark card information stored in the trademark memory, and obtain the card card information and card matching information of the relevant result trademarks; according to the preset trademark similarity rate, The calculation formulas of trademark meaning similarity rate, trademark sound similarity rate and search keyword matching score rate are calculated; the comprehensive quantitative value of trademark similarity is calculated and obtained, and then the resultant trademarks are sorted by the comprehensive quantitative value of trademark similarity.

The above-mentioned Chinese patent (publication number CN107330109A) still has problems that need to be improved. This case mainly focuses on the technical development of trademark search logic. It analyzes, compares and calculates the trademark logo input by the user from different angles such as type, sound and meaning. Similarity and ranking, even if the user is sorted and many previous cases are listed, the user still does not know whether to apply. In addition, the previous cases for the trademark are established in the same category or even the same "similar group" It is based on the concept that even if your logo is similar to someone else’s, but you are selling food and the other person is selling mobile phones, in this situation, someone else’s logo should not affect your trademark application. This case is just a large-scale Search and sort out the degree of similarity. In fact, the applicant still does not know which category of the application is more suitable, and there is still no way to solve related problems such as cross-border trademark recommendation.

As disclosed in U.S. Patent Publication No. US20140280104A1, for example, in one example of the solution described herein, a trademark search is performed to compare the search trademark to a number of potentially relevant references, such as existing trademark registrations, common trademarks, etc. Legal references, domain names, etc. The search results are compiled into a data set that can be stored in any suitable data storage mechanism. This dataset is analyzed to determine proximity scores for multiple categories associated with each reference. For example, these categories may include: similarity in appearance, similarity in sound, similarity in connotation, similarity in commercial impression, similarity in goods/services, similarity in transaction channels, similarity in sales conditions, and the popularity of previous trademarks, etc. The associated proximity score is an objective measure of potential trademark risk and conflict. For example, in one example, the proximity score may be represented by a number from 0-5, where a higher score indicates a higher level of concern. In such an example, a higher proximity score for appearance similarity indicates appearance similarity. Higher, a higher score for a previous mark indicates that the mark is more famous. In other words, the higher the proximity score, the more likely the factor is to indicate a potential violation or conflict.

Although the above-mentioned US patent (Publication No. US20140280104A1) displays the degree of similarity through scoring after comparison and displays through the interface which categories the similar previous patents fall into, there are still problems that need to be improved. For example, users can know which categories to avoid applying for. , but I don’t know which categories are suitable for application. If I want to recommend a cross-border trademark, I can’t help. And the user needs to enter accurate and accurate text in the corresponding field, such as trademark name and product field, in order to effectively proceed with the follow-up. The comparison and scoring are also very limited in actual use.

As disclosed in US Patent Publication No. US20170322983A1, a special UI is used to indicate the degree of similarity between the previous case and the searched trademark. The "display" is to display the similarity of the previous case to the current case through the position of the previous case displayed on different monitors ( (Approximate threat) level, users can observe content classifications of various degrees of "similarity to previous cases" through different block positions of the display.

The trademark search logic and functions of the above-mentioned U.S. patent (publication number US20170322983A1) are limited to the approximate comparison and fraction conversion of "word trademarks". The difference from this case lies in the starting point. This case is based on the recommended recommendations for trademark application classification as the starting point for the previous case. Search and risk assessment, and category recommendation, the purpose of the previous case is to conduct highly professional trademark search-oriented, and through the "monitor" screen presentation and operation, users can classify and view similar trademarks to different degrees. .

Another example is the trademark category search system on the official website of the Intellectual Property Office of Singapore (IPOS). Please refer to Figures 6 to 7. It allows users to enter the type name of goods and services, such as car repair. The system will search for the entered text in the database. Compare and find all product items containing car or repair, display all product items containing car or repair in the search results, and indicate which category they belong to.

The above-mentioned trademark category search system of IPOS is only a simple full-text comparison. If one or two words are different, the search cannot be complete. All included product items will be listed in the process, which does not reach the level of recommendation or even for users. The effect of risk assessment. Furthermore, the text entered by the user must also be a clear product word or service word, otherwise the system cannot perform the comparison. For example, if I run a car repair store is entered, the search result is 0. After entering It is also difficult for ordinary applicants to easily know the trademark category of the application, because ordinary applicants do not necessarily know the name of the trademarked product item or which words are suitable for the search subject.

From the above description, it can be understood that it is necessary to improve or adjust the conventional technology to enhance the convenience of use. In view of this, the inventor of the present invention worked hard on research and creation, and finally developed the system of the present invention.

The purpose of the present invention is to provide an online application system with trademark category recommendation, which solves the above-mentioned problems existing in the prior art.

Therefore, in order to achieve the above-mentioned purpose of the present invention, the present invention provides the semantic analysis trademark category recommendation system, which includes: an electronic device operated by a user, and the electronic device is connected to a server through network information, wherein the electronic device Containing a processor, a memory and a network interface controller, the server includes an application program.

The processor connects to the server through the network interface controller and executes the application, thereby configuring and activating the trademark online application module and the category recommendation module. Furthermore, the processor configures and activates the application data collection unit, category item selection unit, Data transmission unit, text analysis unit, category item comparison unit and report generation unit.

In addition, the processor is also configured to enable a data sending and receiving module to load data from multiple databases into the memory.

The processor further connects to multiple delivery servers through the network interface controller, so that the data transmission unit can upload the user's trademark application documents to the delivery server in the selected country.

The user operates the processor of the electronic device to execute the trademark online application module of the application to apply for a trademark online. The user inputs the information required for trademark application through the electronic device, such as: trademark name, trademark pattern, applicant, contact person, The display screen of the electronic device will display the corresponding fields for the product items to be applied for or the description of the goods and services for the user to fill in the corresponding information. The processor then activates the application data collection unit to input the corresponding information to the user. The data in the fields are classified and stored in the memory. Since the information in the fields of the trademark name, trademark pattern, product items to be applied for, and description of the goods and services in the trademark application data will be particularly used later, the information in these fields is first processed through the application data. Collect units for classification.

Preferably, the processor is configured to enable the category item selection unit to provide the user with the trademark category items that have been pre-stored in the memory for selection through the electronic device.

Preferably, the processor is configured to activate the data transmission unit, connect to a delivery server in at least one country through the network interface controller, and upload the trademark application document completed by the user.

Preferably, the processor is configured to enable the text analysis unit to perform semantic analysis on the trademark application information input by the user to extract keywords.

Preferably, the processor is configured to activate the category item comparison unit, compare the keywords retrieved by the text analysis unit with the trademark category items pre-stored in the memory, and store the comparison results back in this memory.

Preferably, the processor is configured to enable the report generation unit to arrange the comparison results in the memory according to word similarity to generate the category recommendation report.

Preferably, the processor connects to the server through the network interface controller and executes the application, thereby further configuring and activating a risk assessment module to perform pre-processing of the trademark name in the trademark application data input by the user. case comparison, and the processor executes the risk assessment module in the application to further configure and enable the text parsing unit, a search comparison unit and the report generation unit.

Preferably, the processor is configured to enable the text parsing unit to perform text arrangement and combination of the trademark names of the trademark application data input by the user and store them in the memory. The processor is further configured to enable the search comparison unit to All text combinations in the memory are compared with the trademark records pre-stored in the memory, and the comparison results are generated and stored back in the memory.

Preferably, the processor is configured to enable the report generation unit to arrange the comparison results in the memory according to word similarity to generate the risk assessment report.

The following is a detailed description only through specific embodiments and drawings.

Inventive concepts will now be elucidated more fully hereinafter with reference to the accompanying drawings, in which exemplary embodiments of the inventive concepts are shown. The advantages and features of the inventive concept, as well as the methods for achieving them, will be apparent from the following exemplary embodiments, which are explained in more detail with reference to the accompanying drawings.

The terminology used herein is for describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may be present.

Example embodiments are described herein with reference to the drawings, which are idealized illustrations of the examples. Therefore, deviations from the shapes illustrated are expected to occur due, for example, to manufacturing techniques and/or tolerances. Accordingly, the regions shown in the figures are schematic and their shapes are not intended to illustrate the actual shape of the device or to limit the scope of the exemplary embodiments.

Please refer to Figures 8 and 9 in conjunction. Figure 8 is a schematic diagram showing the architecture of the system of the present invention; Figure 9 is a schematic diagram showing another architecture of the system of the present invention. The system of the present invention is implemented by providing an electronic device 100 operated by a user U. The electronic device 100 is connected to a server 300 through the network 200. The electronic device 100 includes a processor 110, a memory 120 and A network interface controller 130, the server 300 includes an application program 310, so that the processor 110 can connect to the server 300 through the network interface controller 130 and execute the application program 310, thereby configuring and enabling the trademark online application module 311 and category recommendation module 312. Further, the processor 110 is configured to enable an application data collection unit 3111, a category item selection unit 3112, a data transmission unit 3113, a text analysis unit 3121, a category item comparison unit 3122, and a report generation unit 3123. In addition, , the processor 110 is also configured to enable a data transceiver module 313 to load the data in the plurality of databases 400 into the memory 120 .

The server can be a cloud server or configured as an on-premises server.

In this embodiment, the author uses a server with the following specifications to perform the training and calculation execution of the trademark product item conversion model of this technology; the processor (CPU) uses a high-performance multi-core processor, especially for processing large amounts of data and performing For complex calculations, use at least a CPU with 16 cores or more (such as AMD Ryzen Threadripper or Intel Xeon series). And the size of the memory (RAM) can be but is not limited to a RAM specification of 64GB or higher to be able to process the size of the corpus and the size of the word vector model. The network interface controller is a high-speed and stable network connection hardware, especially used in cloud computing resources or downloading/uploading large amounts of data. Among them, the most important graphics processing unit (GPU) uses high-performance GPUs (such as NVIDIA's RTX 30 series or Tesla series) to reduce model training time; among them, in the training architecture of the server in this case, the above-mentioned A high-standard training server is used, but when the model training is completed and inference processing is performed, a lower-standard server host can be used, and it can be the same cloud or local server host deployed in the system. In the framework of this case, the adoption of server host specifications does not affect the technical features emphasized in this case, so any server host specifications should still fall within the technical scope of this case. Among them, the processor 100 is further connected to multiple delivery servers 500 through the network interface controller 130, so that the data transmission unit 3113 can upload the trademark application documents of the user U to the delivery servers 500 in the selected country.

Specifically, the user U operates the processor 110 of the electronic device 100 to execute the trademark online application module 311 of the application program 310 to apply for a trademark online. The user U inputs the information required for trademark application through the electronic device 100, such as the trademark name. , trademark pattern, applicant, contact person, product items to be applied for, or description of goods and services, etc., the display screen of the electronic device 100 will display corresponding fields for the user to fill in the corresponding information. The processor 110 Then the application data collection unit 3111 is activated to classify the data input by the user U in the corresponding fields and store it in the memory 120, because the trademark name, trademark pattern, product items to be applied for and the product items to be applied for in the trademark application data will be particularly used later. The information in these fields of product and service description is first classified through the application data collection unit 3111.

Specifically, the processor 110 executes the application program 310 to activate the category item selection unit 3112 of the trademark online application module 311. The user U can directly select the product item for which a trademark is to be applied for through the electronic device 100. The processor 110 accesses the memory 120. The pre-stored list of trademark categories and items is displayed on the display screen of the electronic device 100, allowing experienced or professional users to directly search for category items corresponding to goods or services, and then complete the online trademark application after selection.

If the user U is not an experienced or professional person, he can select the category recommendation option through the display screen of the electronic device 100, and the processor 100 activates the category recommendation module 312 to recommend trademark product items, so that the user U can quickly select by referring to the recommended items. Appropriate category items.

The processor 110 further activates the text parsing unit 3121 to read the trademark application data classified by the application data collection unit 3111 in the memory 120, especially the product items and descriptions of the product services, and the product services input by the user U The text of the description is semantically analyzed to extract keywords. The description of the product and service can be, for example: I want to open a cafe selling hand-brewed coffee, sell drinks, cakes, hand-made biscuits and accessories, etc.

Specifically, keyword extraction is a natural language processing technology that aims to automatically extract important keywords or phrases from text. The method can be but is not limited to statistical methods, frequency-based methods, and text statistics methods. (Statistical methods), text vectorization methods (Text vectorization methods) or machine learning methods (Machine learning methods).

The frequency method determines the importance of words based on their frequency in the text. Common methods include TF-IDF (term frequency-inverse document frequency) and term frequency (Term Frequency). TF-IDF considers the frequency of words in the text and their importance in the entire corpus, while word frequency only considers the frequency of words in the text.

Text statistics method is based on statistical models to analyze the distribution and correlation of words in text. Common methods include mutual information (Mutual Information), pointwise mutual information (Pointwise Mutual Information), and Chi-squared Test (Chi-squared Test). These methods usually require building a statistical model between words and text and calculating the importance of words based on this model.

The text vectorization method converts text into vector representation, and then uses a vector space model to calculate the importance of words. Common methods include Bag-of-Words Model, Word Embeddings, and text vectorization methods (such as TF-IDF vectorization).

Machine learning methods use machine learning algorithms to train models to learn the importance of words from text. Common methods include text classification, text clustering, and keyword extraction models. These methods require the use of annotated text data for model training.

The above can also be executed by writing a program using a scripting language (such as Python).

After the text analysis unit 3121 extracts the keywords, the processor 110 then activates the category item comparison unit 3122 to compare the product items input by the user U in the memory 120, the extracted keywords and the trademark category items. The trademark category items have been connected to multiple databases 400 through the data transceiver module 313 in advance and the trademark category items of different countries are stored in the memory 120. According to the country where the user U wants to apply, the trademark category items of the country are selected and compared.

Specifically, the comparison method can be, but is not limited to, writing a program using a script language (such as Python), reading in each category name one by one, and then reading the trademark category item text in the memory 120 through the program to search whether There are matching or similar items, or you can use SQL language for comparison, and use the SELECT statement to select category items that meet specific conditions to find out whether they exist in the trademark category items.

If you use a scripting language (such as Python) to write a program, you can read each trademark category item one by one, and then use the file processing function in Python to read the text in the trademark category item to search for matching or similar words. In order to carry out For text comparison, you can use string processing methods in Python, such as string comparison, regular expressions, etc. Through these methods, the program can determine whether the text in the trademark category item contains words that match the product item or keyword to determine whether matching text exists.

In addition, the comparison method can be but is not limited to string matching algorithms (such as Levenshtein Distance, Jaro-Winkler Distance, etc.). Levenshtein Distance (also known as Edit Distance) is an algorithm that measures the similarity of two strings. The principle is to calculate the minimum number of operations required to convert one string to another. These operations can be to insert a character, delete a character, or replace a character.

For example, suppose there are two strings "kitten" and "sitting". To calculate the Levenshtein Distance between them, we convert "kitten" to "sitting" and need to go through the following steps: Replace "k" is "s", it becomes "sitten"; replace "e" with "i", it becomes "sittin"; insert a "g", it becomes "sitting", therefore, the Levenshtein Distance between the two strings is 3 , that is, it takes 3 operations to convert one string to another. Through this algorithm, we can calculate the distance between any two strings and use it to compare the similarity between them.

In addition, Jaro-Winkler Distance is an algorithm that measures the similarity of two strings. It is an improved version of Jaro Distance and is usually used to compare shorter strings, such as names, addresses, etc. Jaro-Winkler Distance will measure the similarity between two strings based on The distance between the same character and the character position of each string is calculated as a similarity value between 0 and 1, where 0 means completely dissimilar and 1 means exactly the same. The main steps of the Jaro-Winkler Distance algorithm are as follows: Calculate the Jaro distance and calculate the similarity between two strings. It mainly considers the order of characters in the strings and the number of characters in the two strings; calculate the Winkler correction Factor, mainly used to deal with similar strings. It adjusts the value of Jaro distance by calculating the length of the common prefix and the value of a constant p; calculates Jaro-Winkler Distance, Jaro-Winkler Distance = Jaro distance + Winkler correction factor .

After the comparison results are generated by the category item comparison unit 3122, they are stored in the memory 120. At the same time, the processor 110 activates the report generation unit 3123 to arrange the comparison results according to the degree of word similarity, with identical or high similarity being prioritized. Recommended category items, specifically, the degree of approximation can be calculated in different ways during the comparison process, such as Edit Distance, Cosine Similarity or Jaccard Similarity. After approximation After sorting by degree, a recommendation report is generated. Through the recommendation report, the user U can clearly know how to select trademark category items.

Further, the processor 110 connects to the server 300 through the network interface controller 130 and executes the application program 310, thereby configuring and enabling a risk assessment module 314 to evaluate the trademark in the trademark application data input by the user U. The name is compared with previous records, and the processor 110 executes the risk assessment module 314 in the application 310 to further configure and enable a text parsing unit 3121, a search comparison unit 3142, and a report generation unit 3123.

The text analysis unit 3121 performs text arrangement and combination on the trademark names in the trademark application materials stored in the memory 120. Specifically, if you want to arrange and combine a word into a single text, you can use the following steps: First, put the words into a single text combination. The first character of the word is taken out as the starting character; then, the remaining characters are arranged and combined. This can be done through recursion or iteration. Recursion is a self-invoking process that decomposes the problem into Smaller sub-problems are processed recursively until the termination condition is reached. Iteration uses loops to repeatedly perform the same operation until the termination condition is reached; inserting the first character into different positions of each permutation and combination, forming New permutations and insertion positions can be tried one by one from beginning to end, so as to ensure that all possible permutations and combinations are generated; repeat the above steps until all characters have been processed.

It should be noted that the above permutation and combination methods are only examples and are not limited. Their main function is to facilitate the subsequent comparison process to find similar names as much as possible.

After the text analysis unit 3121 completes storing all the permutations and combinations of the brand name input by the user U into the memory 120, and the processor 110 simultaneously activates the search comparison unit 3142, which has been connected through the data sending and receiving module 313 before The database 400 stores all applied trademark information in the memory 120. These applied trademark information are trademark precedents and serve as the basis for subsequent comparisons.

Specifically, the retrieval comparison unit 3142 compares all the permutations and combinations in the memory 120 with the trademark records in the memory 120. The comparison method is similar to the above-mentioned category item comparison unit 3122, and therefore will not be described again.

The comparison results generated by the retrieval and comparison unit 3142 will also be stored back in the memory 120. At the same time, the processor 110 activates the report generation unit 3123 to arrange the comparison results by degree of similarity. The calculation method of the degree of similarity is the same as the above category recommendation. In the report, the higher the degree of similarity, the higher the ranking order, and a risk assessment report is generated. The report can be presented as an online image, a downloadable document file, a file format that can be forwarded to communication software, or can be shared with the community. Software file format, electronic bulletin board or combination thereof.

In another embodiment, the processor 110 can enable the category recommendation module 312 and the risk assessment module 314 at the same time, and the report generation unit 3123 generates a comprehensive report that combines trademark category recommendation and risk assessment.

Specifically, the text analysis unit 3121 simultaneously accesses the brand name, product items and product service descriptions input by the user U in the memory 120, arranges and combines text on the brand name, extracts keywords from the product and service descriptions, and stores The processor 110 then activates the category item comparison unit 3122 and the retrieval comparison unit 3142 to perform corresponding comparisons between the keywords and all text permutations and combinations, and the generated comparison results are stored back in the memory 120 .

Next, the processor 110 activates the report generation unit 3123 to read the comparison results in the memory 120, and arranges the similarity between the trademark category items and the trademark names. Specifically, the similarity of the trademark names is first arranged, and then the trademark is analyzed and calculated. The degree of similarity between the category items of the trademark and the recommended trademark category items. If the trademark name is highly similar and the category item of the trademark belongs to the highly recommended trademark category item, it means that the previous trademark case is a highly similar previous case, and it will be clearly marked in the comprehensive report. Color is presented.

The colors displayed in the comprehensive report are used to clearly distinguish the trademark category items that the user U tries to avoid. This trademark category item has many highly similar trademark cases, providing the user U with a clear choice of a suitable and low-risk category. project.

Please refer to Figure 10 for another embodiment of the present invention. The system is used to receive a string provided by a user by operating an electronic device and at least one target country to be queried, through a processor of the electronic device. A network interface controller is connected to a server and executes an application program to calculate and generate a trademark category recommendation report. The system at least includes: an input module 600, a multilingual translation model 700, a category recommendation module 312, and a risk assessment module. Group 314 and database 400, where the database 400 can be a Nice trademark category database, a trademark category database of multiple countries, or a multinational trademark database.

Among them, the Nice trademark category database and the trademark category databases of multiple countries mainly store the classification and product item data of the Nice international trademark category, as well as the trademark category classification and product item data of each corresponding country. The multi-country trademark data The database stores trademark application data for each corresponding country.

Users can choose from any combination of brand owners, trademark applicants, trademark practitioners, and legal practitioners.

The input module 600 is used to receive a string input by the user, perform tokenization processing on the string, and send a string of information. Specifically, string tokenization (Tokenization) is to decompose a sentence or document. The process of dividing into individual words (tokens), here is a simple string labeling processing method:

Remove punctuation marks: Use regular expressions or a predefined list of punctuation marks to remove punctuation marks from the string, such as periods, commas, question marks, etc. Split words: Split the string according to spaces or other specific delimiters, and each split part is a word. Handle special cases: Handle special cases such as abbreviations, hyphens, numbers, etc. Regular expressions or specific rules can be used to handle these situations, splitting them into appropriate words. Convert to lowercase (if it is a case-sensitive language): Convert all words to lowercase to unify the representation of words.

The multilingual translation model 700 is a computing model trained through large-scale language translation. When receiving the string information, it determines whether the input string language is consistent with the official language of the target country selected by the user. If it is not consistent, the The string entered by the user is translated, the string is translated into the official language of the target country, and the translated string information is sent.

The category recommendation module 312 is a computational model trained through natural language models and trademark classification tables and details to parse ambiguous semantic (or imprecise description) string information into trademark category recommendation information, which also includes: text Analysis unit 3121, category item comparison unit 3122, and report generation unit 3123.

Natural language models can predict the next word or generate sentences that conform to grammar and semantics based on known text data. They can be but are not limited to:

N-gram model: The N-gram model is a language model based on probability. It assumes that the probability of a word appearing is only related to the previous N-1 words. For example, in a bigram model, predict the probability of the next word given the previous word.

Recurrent Neural Network (RNN) model: RNN is a neural network suitable for processing sequence data, which can capture the temporal dependence between words. In natural language processing, RNN is often used in the construction of language models, where each word is regarded as a time step.

Pre-trained language model (such as BERT): A pre-trained language model is a model obtained through large-scale unsupervised training that can understand and generate natural language. The BERT model uses the Transformer network architecture to be pre-trained on a large amount of text data, and then fine-tuned to adapt to specific tasks, such as text classification, named entity recognition, etc.

The trademark classification table and details are the classifications extracted from the trademark classification databases of various countries, as well as all the product items in the classifications. They are extracted and formed into a list, and a natural language model is provided for training and calculation.

The text parsing unit 3121 is used to parse the translated string information and extract keywords and descriptive text collectively referred to as industrial information such as business activities, products, services, etc.

The category item comparison unit 3122 is used to receive the relevant description text of the keywords and industry information, compare it with the trademark category database of the target country, and calculate to generate at least one category information that is closest to the target country. One category of product item information and one category of service item information are recommended categories and recommended product items, and one recommended category is matched with at least one recommended product item.

The report generation unit 3123 is used to receive the recommended category, the recommended product item and the target country, and use the input language to generate a trademark category recommendation report through a templated format. Specifically, if the word input by the initial user If the language of the string is different from the official language of the target country, the generated trademark category recommendation report will then use the multilingual translation model 700 to translate the recommended category and recommended product items back to the language used when the string was input.

Specifically, the input string may include trademark text, trademark description, business behavior, company name, stock code, product name, service name or combination information thereof.

The method of inputting strings on the user side can be text input, voice input or video input.

The input module 600 further includes an information retrieval unit 601, which is used to actively retrieve the tagged string information to the network space to retrieve relevant information, and can transmit the information to the multilingual translation model 700 for translation.

Specifically, the information retrieval unit 601 crawls information in the cyberspace based on the industry-related information generated after the text string input by the user is parsed by the text parsing unit 3121. The information crawled here can be For the user's official website, news or other online articles containing the extracted keywords.

The risk assessment module 314 is used to compare the keywords extracted from the input string with the trademark database of the target country, and generate a risk information. The risk information can also be generated through the report generation unit. 3123 is integrated into the trademark category recommendation report to generate a risk assessment report.

Specifically, the recommended trademark categories and recommended product items are first generated through calculation by the category recommendation module 312, and then the risk assessment module is used to search and compare these recommended categories. First, the keywords are compared with the trademark information of the target country. Whether there is a similar name in the library, and if there is a similar previous name in the recommended category, then compare the similarity of the product items, and finally generate a risk assessment report after cross analysis.

Please continue to refer to Figure 11. The category recommendation module 312 of this system further includes a knowledge graph comparison unit 3124. Specifically, the information retrieval unit 601 can not only crawl the user's own official website or related network information, but also Based on the parsed industrial information, company information of similar industries or similar backgrounds can be crawled, and then the information can be compared through the knowledge graph comparison unit 3124 to compare previous trademark cases applied by companies in similar industries or similar backgrounds. The categories applied for in these cases and the product items are retrieved and merged into the recommended categories and recommended product items compared by the category item comparison unit 3122.

Knowledge Graph (Knowledge Graph), in which each node represents the trademark application case record data of different companies, and the edges represent the industrial correlation or similarity between the companies. This method can visually display the relationship between categories/product items and industries in trademarks applied for by different companies, and the construction of this knowledge map is collected over time, which includes but is not limited to the following information : Completed registration trademark announcement information (trademark family), a comparison table of previous cases constructed through word meaning expansion and translation comparison, and some trademark application cases disclosed by the National Intellectual Property Administration.

Moreover, its detailed construction process is as follows:

Data collection: Collect existing trademark information, including trademarks, trademark names, categories and other related information that the company has applied for. This information can be obtained from sources such as trademark databases and patent offices.

Knowledge graph construction: Combine trademark information with other related information, such as company information, industry categories, etc., to construct a knowledge graph. Knowledge graphs can use graph databases or graphical representations to represent relationships between different entities (e.g. companies, trademarks, industries).

Similarity calculation: Calculate the similarity between different trademarks based on the relationships and attributes in the knowledge graph. This can be based on different characteristics, such as similarity of the brand name, similarity of the category it belongs to, etc.

For another embodiment of the present invention, please refer to FIG. 12 , which shows a flow chart of the method of the present invention, which includes steps S101 to S108.

In step S101, the user inputs the introduction text, mainly by operating the electronic device 100 to input any text, which is not limited to trademark names or standard words; in step S102, the text analysis unit 3121 performs semantic analysis on the introduction text input by the user. Analyze and determine whether the input text contains the user's brand name, company name, company introduction, product introduction and other information; if so, proceed to step S103, the text analysis unit 3121 compares the brand name and company name in the introduction text input by the user Extract it; and then proceed to step S104 to determine the identification of the extracted text. This is mainly because one of the approval conditions for a trademark is that it must be identifiable. In step S105, the text analysis unit 3121 is used to first exclude non-identifiable characters. text and extract the identifiable text. For example, the "Incorporated Company" in ABC Co., Ltd. is non-identifiable text; if the aforementioned judgment is no, step S106 is performed after step S105, and the introduction text or specific information input by the user is The identifying text is again analyzed through the text analysis unit 3121 for industry analysis and product and service content analysis. In step S107, the category item comparison unit 3122 is also activated to parse and analyze the identifying text in step S106. The introduction text is compared with the trademark category items in the database 400, and in step S108, the comparison result is used to form a category recommendation report through the report generation unit 3123, and the recommendation reasons are further attached.

Another embodiment of the present invention differs from the embodiment of FIG. 12 in that after step S106, in step S106, the introductory text or identifying text input by the user is again analyzed through the text analysis unit 3121, and the product is also analyzed. and service content analysis. Furthermore, the category item comparison unit 3122 first compares the analyzed industry descriptions, and then generates recommended trademark categories. Then, the category item comparison unit 3122 compares the analyzed product and service content. Describes comparing the already generated recommended trademark categories, and then generating recommended product items of the trademark category, and finally proceeding to step S108, where the comparison results are formed into a category recommendation report through the report generation unit 3123, and the reasons for the recommendation are further attached. .

Please continue to refer to FIG. 13 , which shows a method flow chart of another embodiment of the present invention, which includes steps S201 to S208.

In this embodiment, in step S201, the user copies and pastes the text description of the company introduction information by operating the electronic device 100. In particular, what the user inputs in step S201 is a text description and not just words or nouns; in step S202 , the text analysis unit 3121 performs semantic analysis on the text input by the user, extracts keywords; and determines whether the input text description contains the user's brand name, company name and other information, and the subsequent steps S203 to S208 are the same as the aforementioned step S103 ~S108 is the same, so it will not be described again.

Please continue to refer to FIG. 14 , which shows a method flow chart of another embodiment of the present invention, which includes steps S301 to S309.

In step S301, the user inputs introduction text by operating the electronic device 100, and determines through the text analysis unit 3121 whether the input language is the default received language and whether translation needs to be performed; if translation is required, step S302 is first performed for language translation. If There is no need to translate or complete step S302 before proceeding to step S303. The text analysis unit 3121 performs semantic analysis on the introduction text input by the user and extracts keywords; and subsequent steps S304 to S309 are the same as the aforementioned steps S103 to S108, so no further steps are required. Repeat.

Please continue to refer to FIG. 15 , which shows a method flow chart of another embodiment of the present invention, which includes steps S401 to S408.

In step S401, the user narrates and inputs descriptive text by operating the electronic device 100, and converts the speech into text through the text analysis unit 3121 and stores it in the memory 120. The speech recognition technology can be based on models and algorithms to convert the audio data into text. Compare with the speech model to identify the corresponding text content, and perform post-processing of the text, such as grammar correction, spelling correction and semantic analysis; then judge the text to determine whether it contains the user's brand name, company name and other information; The subsequent steps S403 to S408 are the same as the aforementioned steps S103 to S108, and therefore will not be described again.

Please continue to refer to FIG. 16 , which shows a method flow chart of another embodiment of the present invention, which includes steps S501 to S509.

Steps S501 to S507 in this embodiment are the same as the aforementioned steps S101 to S107, and therefore will not be described again. In step S508, the search comparison unit 3142 is enabled to compare the retrieved identifying words with the applied trademark records in the database 400, and store the comparison results in the memory 120; in step 509 , the report generation unit 3123 sorts the comparison results to generate a category recommendation report and a list of similar cases, and at the same time attaches the reasons for the category recommendation.

Please continue to refer to FIG. 17 , which shows a method flow chart of another embodiment of the present invention, which includes steps S601 to S611.

In this embodiment, the main purpose is to provide users with trademark category recommendations from different countries (cross-border countries). Steps S601 to S608 are the same as the aforementioned steps S101 to S108, and therefore will not be described again. After step S608, the display screen of the electronic device 100 displays whether to perform cross-border commodity item conversion to provide the user with a choice. If not, the end is completed. If so, step S609 is performed to provide multiple countries in the database 400 for the user to choose. Different from the original The country of application, and then select at least one second application country; after the user selects the country in step S610, the processor 110 reads the trademark category items of the country in the memory 120 and activates the category item comparison unit 3122 for comparison; In step S611, the comparison result is displayed on the display screen of the electronic device 100, thereby achieving a cross-border trademark category recommendation function.

Please continue to refer to FIG. 18 , which shows a method flow chart of another embodiment of the present invention, which includes steps S701 to S709.

In this embodiment, in step S701, after the user inputs the introduction text by operating the electronic device 100, he first selects the country for application. After determining the country for trademark application, step S702 is performed. The text analysis unit 3121 performs semantic analysis on the introduction text input by the user. And determine whether the introduction text contains information such as brand name, company name, etc., and subsequent steps S703 to S707 are the same as the aforementioned steps S103 to S107, so they will not be described again. In step S708, the comparison results generated in step S707 are converted across countries according to the country selected by the user in step S701, and the original recommended category items are converted into product items of the selected country; in step S709, the report generation unit 3123 The conversion results are output to generate a cross-border category recommendation report.

Please continue to refer to FIG. 19 , which shows a method flow chart of another embodiment of the present invention, which includes steps S801 to S813.

In this embodiment, category recommendations and risk assessments that can be converted across countries are also provided, and steps S801 to S806 are the same as the aforementioned steps S101 to S106, so they will not be described again. In step S807, the extracted identifiable text and the parsed text are language converted through the text parsing unit 3121. In this embodiment, all text is preset to be uniformly converted into English; in step S808, the category item ratio is The comparison unit 3122 compares the text that has been converted into English with the information data that has been converted into English in advance in the database 400; in steps S809 and S810, it compares the product items and the applied case names, respectively. The applied case names are compared through the search comparison unit 3142; in step S811, the comparison results of steps S809 and S810 are comprehensively analyzed by the report generation unit 3123; in step S812, the report generation unit 3123 performs the comparison The priority of category recommendations is distinguished by color in the results; in step S813, the report generation unit 3123 comprehensively analyzes the category recommendations and search results and then integrates and generates a transnational risk assessment report.

Please continue to refer to FIG. 20 , which shows a method flow chart of another embodiment of the present invention, which includes steps S901 to S914.

This embodiment also provides a cross-border category recommendation and risk assessment method, and steps S901 to S906 are the same as the aforementioned steps S101 to S106, and step S907 is the same as the aforementioned step S807, so they will not be described again. In step S908, the text parsing unit 3121 is enabled to generate synonyms or similar words between the extracted identifying text and the parsed text. The method of generating similar words or synonyms may be, but is not limited to, the text parsing unit 3121 using existing vocabulary. Table, include synonyms or similar words of words into the vocabulary list, and generate similar words or synonyms by matching the words in the vocabulary list. For example, WordNet is a synonym generation system based on the vocabulary expansion method, or uses a large number of corpora to learn the relationship between words, including word co-occurrence and contextual similarity, etc., to generate similarity scores between words, and then generate approximations words or synonyms. For example, LSI (Latent Semantic Indexing) and LDA (Latent Dirichlet Allocation).

WordNet is a computerized vocabulary database of English vocabulary, which contains a large number of English words and uses the semantic and grammatical relationships of words as a basis to organize and manage words. The purpose of WordNet is to provide a reliable language resource for natural language processing and semantics. analyze.

LSI (Latent Semantic Indexing) is a natural language processing technology used to identify synonyms and similar words. Its basic principle is to identify words by converting text into a vector space model and then performing Singular Value Decomposition (SVD) Specifically, LSI uses word frequency statistics in the text to process the words in the text and convert them into a vector space model. Before performing SVD, LSI usually uses TF-IDF (Term Frequency-Inverse Document Frequency) weights to weight words to eliminate the influence of some commonly used words, thereby better capturing the correlation between words. Through this processing, LSI A word-text matrix can be generated in which each word corresponds to a vector. Through singular value decomposition, LSI can decompose the singular values and singular vectors of this vocabulary-text matrix, thereby capturing the semantic information implicit in the text. Based on this semantic information, LSI can calculate the similarity between two words to find synonyms or similar words.

In addition, LDA (Latent Dirichlet Allocation) is a commonly used topic model that can allocate texts in a collection to multiple topics, and at the same time find the vocabulary distribution represented by each topic. This vocabulary distribution can be used to find synonyms or To approximate words, you first need to prepare a text set. This text set can be any text set that contains the target vocabulary. Then, use LDA to distribute this text set into multiple topics. At the same time, you can find the vocabulary distribution represented by each topic. For a target vocabulary, you can use its corresponding vocabulary distribution in the LDA model to find synonyms or similar words. Specifically, the similarity between the vocabulary distribution of the target vocabulary and the vocabulary distribution of other vocabulary can be calculated, and vocabulary similar to the target vocabulary can be found as synonyms or approximate words.

The text parsing unit 3121 can also use a script language (such as Python) to write a program to execute WordNet, LSI (Latent Semantic Indexing), and LDA (Latent Dirichlet Allocation) to perform text parsing and generate synonyms or similar words.

Among them, it is worth mentioning that the machine translation in the technology of the present invention adopts a direct method and a machine translation model based on deep learning, such as Google's Transformer model or Facebook's FAIR SEQ model, to translate a country's trademark product item sentences into Direct translation into another country's language. Thereby entering another country's commodity item database for direct text comparison.

In addition, a knowledge graph (Knowledge Graph) of multinational trademark product items is established in the technology of the present invention. The knowledge graph is used to establish a node matrix association, in which each node represents a product item in a different country, and the edges represent the connections between the product items. Relevance or similarity. Based on the dynamic learning model that evolves over time, the equivalence relationship between product items in various countries and trademark items in each country, the knowledge graph unit conversion at least includes the following steps: Collecting data on the relationship between multi-country trademark product items and trademark projects and using a dynamic model Learning, through graph query and natural and grinding line reasoning, calculate the best match, shortest path, maximum flow, and optimize the equivalence relationship of product items based on the final decision results. This method can visually display the conversion relationship between commodity items in different countries, and the construction of this knowledge map is collected over time, which includes but is not limited to the following information: Transnational trademark announcement information that has completed registration (trademark family), the conversion results completed by previous users, the product item comparison table constructed through word meaning expansion and translation comparison, and some product item translation comparison tables disclosed by the National Intellectual Property Administration.

The detailed construction process is as follows: In the cross-border application conversion technology of trademark goods items (goods or services), the knowledge graph may include the following elements:

Entity: Trademark product items in various countries.

Relationship: similarity or equivalence between various product items. For example, a certain trademark product item in the United States may have the same or similar meaning as a certain product item in Germany. Using the knowledge graph to convert cross-border trademark applications follows the following Steps:

Data collection: First, collect data on trademark product projects in various countries and the relationships between these product projects. This part requires a lot of manual effort, and part also requires automated data capture and processing technology. And use the collected data to build a knowledge graph, in which each product item is a node, and the relationship between each node is an edge.

Graph query and reasoning: Using query and reasoning technology, it is based on the knowledge graph to find the best match between the trademark product project of the original country and the product project of the target country, which involves the shortest path, maximum flow and other issues in graph theory. . The last step is execution result evaluation and feedback: the results of query and reasoning are evaluated. If the results are inaccurate or unsatisfactory, the knowledge graph can be updated based on the feedback.

On the other hand, word vector analysis based on the technology of the present invention is used in the transnational application conversion technology of trademark product items to capture and understand the semantic correlation of product items in trademark classification systems of different countries, and realize the conversion process.

The following provides one of the application methods of the technology of the present invention and the establishment of database models and training methods:

Establish a multilingual word vector model: First, we convert the descriptions of trademarked product items in different countries into word vectors. This is achieved by training a multilingual word vector model. For example, we can use a public large-scale multilingual text data set to train this model, or use an already trained multilingual word vector model, such as Facebook's FastText but not Limit the above language model.

Convert product item descriptions: Convert the original country's trademark product item descriptions into word vectors, and then convert these word vectors into the language of the target country. This step can be achieved through the similarity between word vectors. For example, we can find the nearest neighbors of the word vectors in the original language and the word vectors in the target language, and use the words of these nearest neighbors to form a new product item description.

Manual review and adjustment: Although the word vector model is powerful, it still cannot fully understand the nuances of language and cultural differences. Therefore, after the conversion is completed, this training effect can be enhanced through manual review and correction. For example, a team of experts review and correct the conversion results, and these corrected data are reused to train the model to continuously improve the conversion accuracy of the model. And continue to optimize the model: Over time, the model can be continuously optimized based on actual needs and effects, for example, adding support for more languages, or improving the ability to handle specific types of commodity items.

The above method is only a possible way, and the actual implementation may be adjusted according to specific needs and conditions. Taking the actual conversion case of this case as an example, it is completed through the following "word vector" conversion model, which includes:

Word vector training: Use a large-scale corpus to train a word vector model, and map each word to a high-dimensional space based on context, where similar words will be close to each other in this space. Its transformation formula is the following objective function:

where T is the number of words in the corpus, c is a selected window size, θ is the model parameter, is the context word given the center word w_t conditional probability.

Conversion function: Using the trained word vector model, we can generate corresponding word vectors for product items in country A and country B, and we establish a conversion function to map the word vectors of product items in country A to products in country B. item word vector. And minimize the distance between the word vector of country A and its corresponding word vector of country B. Suppose we have a set of product item pairings (x_i, y_i) from country A and country B, where i = 1, ..., N, N is the number of pairs, then the formula for minimizing the objective is:

where W is the transformation matrix we wish to learn. Assume that the US trademark product item we want to convert is "computer software". We first convert this phrase into a word vector, and using the already trained word vector model, we can get the word vector representation v_US of "computer software". Then, we apply the transformation matrix W to v_US, that is, W * v_US, and the result is the converted word vector v_TW. Next, we need to find the Taiwanese trademark product item closest to v_TW. This can be achieved by calculating the similarity between v_TW and the word vector of the Taiwan trademark product item. The similarity is generally measured using cosine similarity. The calculation formula is as follows: where a and b are two word vectors, a·b represents the vector dot product, ||a|| and ||b|| represent the norms of vectors a and b respectively.

Finally, the system queries the word vectors of all Taiwan trademark product items and finds the item with the highest cosine similarity to v_TW. This item is the converted result. Among them, the system can determine that the most similar Taiwanese trademark product item found is "computer software" and set it as the result of the conversion.

Among them, in the framework of this system and method, ensuring the update of each country's commodity item database is the core focus. Since each country's commodity item database may be updated from time to time once or twice a year, some of the items may be eliminated. Old product item names will also be added with new product items; among them, if the newly added product item is officially notified that it corresponds to a product item of the original old version, such information must be updated in the system at the same time. In the knowledge graph; on the contrary, if the official does not inform the corresponding relationship, the comparison relationship can be maintained through word-image comparison or manual methods.

In step S909, the text in steps S907 and S908 is compared with the language-converted trademark category items in the database 400 through the category item comparison unit 3122; in steps S910 and S911, category recommendations are generated based on analysis of the text. At the same time, the search and comparison unit 3142 performs a previous case search based on the identifying text; in step S912, the report generation unit 3123 conducts a comprehensive analysis of the search results and comparison results, and distinguishes recommended categories by color in step S913, and then in step S913 In S914, the report generating unit 3123 performs a comparison of the search results in the recommended category to generate a risk assessment report.

Next, please refer to FIG. 21A and FIG. 21B, which are schematic diagrams showing application examples of the present invention. As shown in the figure, the user can enter any descriptive text or introductory text into the system's web page (https://inta.aiplux.com/). The text content can include but is not limited to brand name, company name, product description, Service description, company philosophy, company introduction, etc. After entering the description text, you can press the analysis button.

The results are generated immediately after the operation of the category recommendation module of this system. The results include the brand name, company name, and trademark name extracted from the description text, as well as the industry information parsed from the description text. The most important thing is, Recommended trademark categories generated after comparison based on the parsed industrial information, as well as recommended product items in the trademark category.

Next, this case will be explained and displayed through the specific process, and then explain that when the user enters any narrative to enter the system of the present invention, the background operation process of the system is specifically divided into the following five main actions and is implemented using corresponding technologies: Entity Recognition (NER), topic modeling or text classification, recognition text judgment, trademark category recommendation, and trademark product item (goods and services) recommendation. Among them, for example, when the user inputs system information, "AIPLUX Technology Co., Ltd. released Taiwan's first large-scale language model. This model uses up to 176 billion parameters created by supercomputers, combines semantic understanding and text generation capabilities, and launches enterprise-level generative formulas." AI solutions. 』For example, in the entity recognition (NER) stage: the system uses the NER model to identify the organization name in the text (such as "AIPLUX Technology Co., Ltd.").

Moreover, in the topic modeling or text classification stage, it assists the system to understand the main topics in the text and classify them into corresponding industries, and then uses topic modeling to identify "artificial intelligence", "natural language processing" and "big words". "data" and other keywords, and classify them into the corresponding industry categories. In addition, in the recognition text judgment stage, the system needs to find keywords in the text. These words can represent the main name of the company. In this example, the system regards "AIPLUX" as a recognition word.

In the trademark category recommendation stage, the system recommends international categories of trademarks based on the above information. Among them, if a company is mainly engaged in artificial intelligence, natural language processing and big data related businesses, the system will recommend categories 9 (scientific instruments), 42 categories (scientific and technical services) and 41 categories (educational and entertainment services) and other related categories trademark category. Finally, in the trademark product item (goods and services) recommendation stage, the system recommends related goods and services based on the company's industry category and main business. Among them, if a company is mainly engaged in the research and development of artificial intelligence, we can recommend products and services related to artificial intelligence, such as "artificial intelligence consulting services", "artificial intelligence technology research", etc.

In more detail, entity recognition (NER) is a natural language processing (NLP) technology and is used to identify words with specific meanings in text, such as names of people, organizations, places, etc. It is not limited but can be accomplished using the following models: Hidden Markov Model (HMM), Maximum Entropy Markov Model (MEMM), Conditional Random Field (CRF), or neural network-based models such as BiLSTM-CRF. Among them, in this case, it is executed through the BiLSTM-CRF model. The specific steps are to perform word embedding on each word and convert the word into a dense vector representation. Word embedding can use words such as word2vec, GloVe (Global Vectors for Word Representation), CKIP (Chinese Knowledge and Information Processing) or BERT (Bidirectional Encoder Representations from Transformers) and other pre-trained models are obtained. The word embedding is processed using a Bidirectional Long Short-Term Memory (BiLSTM) network, where the representation of each word in its context is obtained at this stage. Then, the output of BiLSTM is sent to the CRF layer: the CRF layer can consider the dependencies between tags, making the sequence labeling results more accurate.

In this display, the user inputs "AIPLUX Technology Co., Ltd. released Taiwan's first large language model for enterprise use. The model uses up to 176 billion parameters created by supercomputers, combines semantic understanding and text generation capabilities, and launches enterprise-level Generative AI solutions. 』For example, the sentence is divided into word sequences: ["AIPLUX Technology Co., Ltd.", "published", "Taiwan", "first", "enterprise use", "big", "language model", " The model", "through", "super", "computer", "build", "out", "of", "Gundam", "1,760", "billion", "parameters", "combination", "traditional "in", "of", "semantic understanding", "and", "text generation capability", "launch", "enterprise level", "generative", "AI", "solution"], further use pre- Trained uses a pretrained word embedding model (such as word2vec, GloVe, or BERT) to convert each word into a vector.

For example, we assume that the word vector of "AIPLUX Technology Co., Ltd." is [0.1, 0.2, ..., 0.5] (in fact, a word vector usually has hundreds to thousands of dimensions), and the word vector of "published" is [0.2, 0.3, ..., 0.6], etc. Feed the word vector into the BiLSTM network. BiLSTM can obtain contextual information for each word. For example, given the word vector of "AIPLUX Technology Co., Ltd.", BiLSTM can generate a new vector based on the words before and after it ("published" and "Taiwan"). This vector contains the words "AIPLUX Technology Co., Ltd." Contextual information in sentences. Feed the output of BiLSTM into the CRF layer. The CRF layer will give the label probability of each word. For example, CRF may determine that the tag of "AIPLUX Technology Co., Ltd." is "ORG" (representing the organization name), the tag of "published" is "O" (representing non-entity words), and so on.

Finally, we select the label sequence with the highest probability as the result of NER. For example, for the above sentence, the result of NER might be: ["ORG", "O", "O", "O", "O", "O", "O", "O", "O", " O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O" , "O", "O", "O", "O", "O", "O", "O"]

Therefore, this result indicates that "AIPLUX Technology Co., Ltd." is recognized as an organization name, and the rest of the words are recognized as non-entity words.

Next, in the topic modeling or text classification stage, topic modeling usually uses unsupervised learning algorithms, such as Latent Dirichlet Allocation (LDA), to identify the topic of the document by analyzing the word frequency of the document. Text classification uses supervised learning algorithms, which require pre-labeled data. The algorithm technologies used in this case include but are not limited to the following algorithms: Naive Bayes (Naive Bayes), Support Vector Machine (SVM) , or deep learning methods such as convolutional neural network (CNN), long short-term memory network (LSTM), etc. In this demonstration, the main technology of the topic modeling LDA algorithm used is that each document can be regarded as a mixture of topics, and each topic can be regarded as a probability distribution of words. In terms of specific implementation, LDA uses Dirichlet distribution to model the distribution of documents to topics and the distribution of topics to words, and then optimizes it through repeated operations to obtain the topic distribution of each document and the word distribution of each topic. After word segmentation, we obtain the term representation of the document. During the LDA calculation process, the system can obtain the term distribution of each topic and the topic distribution of each document. By checking the term distribution of each topic, Find the keywords for each topic. Among them, "artificial intelligence", "natural language processing" and "big data" are determined to be keywords of the same theme, and then excerpted.

In the recognition text judgment stage, keyword extraction can be completed through methods such as TF-IDF or TextRank. Among them, TF-IDF considers word frequency (TF) and inverse document frequency (IDF) to evaluate the importance of a word. The main task of identifying text judgment is to find those parts that are uniquely identifiable to the trademark, that is, those words that can help consumers identify and distinguish the source of the product. In the example you provided, "AIPLUX" may be judged as an identifying word because it may be a specific brand name or company name, while "Technology Co., Ltd." may be judged as a non-identifying word. Because it is a general term, it cannot help consumers distinguish the source of goods. This task can be achieved through the following steps: first, build a dictionary containing all registered trademarks, which can be obtained from the trademark databases of various intellectual property bureaus; second, build a dictionary containing all common words, which can be obtained from Obtained from large-scale text materials, such as Wikipedia or other large corpora. Finally, the above two dictionaries are used to identify the identifying words and non-identifying words in the text.

Among them, there are still shortcomings and risks of judgment distortion simply through the evaluation methods of the above two dictionaries. Recognizable words are usually unique to a certain extent, so they are unlikely to appear in common vocabulary dictionaries. However, this assumption does not always hold, and some identifying words may also be common words. Therefore, in practical applications, we may need to combine other information to improve the accuracy of identifying text judgments, such as using machine learning models to predict whether a word is identifiable, or using knowledge graphs to understand the meaning of a word in a specific context. meaning.

Following the above, in the embodiment of the present invention, further, the system uses the TF-IDF method to extract keywords. It is then used to evaluate the importance of a word in a collection of documents. The calculation formula of TF-IDF is:

Among them: t is a word, d is a document, D is the set of all documents, TF(t, d) is the term frequency of word t in document d, which can usually be calculated by counting the occurrence of word t in document d. times and divided by the total number of words in document d; IDF(t, D) is the inverse document frequency of word t, which can be calculated by dividing the total number of documents by the logarithm of the number of documents containing word t calculate.

For "AIPLUX Technology Co., Ltd.", the system determines that it is a document containing two words. "AIPLUX" appears once and "AIPLUX Technology Co., Ltd." also appears once, so their TF values are both 1/2 = 0.5. Next we need to calculate the IDF value. Assume that our file library contains 10,000 documents. "AIPLUX" only appears in 10 documents, and "Technology Co., Ltd." appears in 1,000 documents, so their IDF values They are log(10000/10) = 3 and log(10000/1000) = 1 respectively. Finally, we multiply the TF value and the IDF value, and the TF-IDF value of "AIPLUX" is 0.5 * 3 = 1.5, while the TF-IDF value of "Technology Co., Ltd." is 0.5 * 1 = 0.5. From this we can see that "AIPLUX" is more important than "Technology Co., Ltd." in this document.

Furthermore, the present invention uses knowledge graph (Knowledge Graph) technology to further optimize the recognition text judgment effect in this technology. The knowledge graph can store a large amount of information about entities and their attributes and relationships. In this problem, we can use "AIPLUX" and "Technology Co., Ltd." as entities. Their attributes can include the number of times they appear in the document, their TF-IDF values, etc., and their relationship can be that they are in the same co-occur in documents, etc. In addition, the knowledge graph in this case integrates and stores data from the intellectual property bureau’s trademark review process. For example, during the review process, "Technology Co., Ltd." or "Technology Co., Ltd." were mentioned many times and were regarded as non-identifying words. Then we can add this information to the knowledge graph, "Technology Co., Ltd." Co., Ltd." can add a "identification: low" tag to the attribute; in this way, in the subsequent keyword extraction process, we can use this information to assign a lower value to the word "Technology Co., Ltd." The weight makes the model more inclined to recognize "AIPLUX" as a keyword. Then we can better understand the entities in the text and their attributes and relationships, so as to more accurately extract keywords and judge identifying text.

In the trademark category recommendation stage, in this stage, the system uses the keywords and topics previously extracted from the text, as well as their corresponding industry classifications, to recommend corresponding trademark categories to users. For example, for a company mainly engaged in artificial intelligence, natural language processing and big data related businesses, the system will recommend related categories 9 (scientific instruments), 42 categories (scientific and technical services) and 41 categories (educational and entertainment services). Trademark category. The process is achieved by consulting data tables and recommendation system algorithms. Among them, the system creates a lookup table in the background to map keywords or topics to their corresponding trademark categories, as shown in the following table: Keyword/topic | Corresponding trademark category ----------------|----------------- Artificial Intelligence | 9, 42 Natural Language Processing | 9, 42, 41 Big Data | 9, 42

Then, the system background can find their corresponding trademark categories based on the keywords or topics that appear in the text entered by the user, and recommend these categories to the user. At the same time, the system's backend system merges the user's industrial information results into the business information query platform to find other companies in the same industry, and then consolidates and summarizes the trademark applications filed by these other companies. , Finally, the recommendation results are compared and analyzed with word vectors, and the recommendation results are compared and analyzed with industry information to integrate the selection suggestions for recommended trademark categories and product items (goods and services).

Among them, the system backend can not only select goods and services through the general Nice classification content, but also select goods and services from each country based on the product information provided by the user. Specifically, the system backend The system provides word vector analysis of product items in various countries, and then conducts word vector analysis and comparison based on the user's introduction information, and finally selects the corresponding recommended "product items" content.

Among them, it mainly includes two stages: one is to use word vector (Word Vector) comparison to recommend goods and services, and the other is to use industry information comparison to further optimize the recommendation results. When the system background already has word vectors for goods and services in various countries, the system can first convert the product names in the user introduction information into word vectors. For example, assume that the product introduced by the user is "AI intelligent robot". During this demonstration, the Word2Vec model is used to convert this word into a word vector v.

Then, the system background can traverse the word vectors of goods and services in various countries, calculate their similarity with v, and use cosine similarity to calculate the similarity between word vectors. The formula is as follows:

where v and w are two word vectors respectively, and "||" represents the length of the vector. Finally, the system can select those goods and services with the highest similarity as recommended results.

On the other hand, in the process of recommending goods and services through industry information comparison, the system background uses the user's industry information to find other companies in the same industry, and then finds the most common goods and services from the trademark applications of these companies. . Specifically, we first search for companies in the same industry on the business information query platform, and then collect the trademark applications of these companies. We can treat the goods and services in each application as a word and calculate the frequency of each word. Then, the system background can select the words with the highest frequency as recommended products and services.

As a result, the system background has currently obtained two recommendation results: one is based on the result of word vector comparison, and the other is based on the result of industry information comparison. Then, the system background can combine these two results to form the final recommendation result. Specifically, each product and service item is assigned a score, which is the weighted average of the item's ranking in the word vector comparison results and its ranking in the industry information comparison results. Then, we can select the item with the highest score as the final recommendation result.

Here is the formula for this step:

Among them, rank_vector(item) is the ranking of the item in the word vector comparison results, rank_industry(item) is the ranking of the item in the industry information comparison results, α and β are the weights of the two methods, and their sum is 1. In practical applications, the values of α and β can be determined through cross-validation to optimize the quality of recommendation results. Therefore, α and β are adjustable parameters. In the introduction of this case, the weight adjustment method and process will not be described in detail. The adjustable parameters are customized according to the user. The adjustable parameters can be selected from the options of user preferences, user habits, and the user's ability to tolerate risks.

If the company name, brand name or trademark name cannot be found in the description text input by the user after being parsed by the system, it will not affect subsequent category recommendations. The system can still parse the description text to generate industry information, and then based on the industry information After comparison, a recommended trademark category is generated, as well as recommended product items in the trademark category.

Please continue to refer to Figure 23 and Figure 24, which are flow charts of embodiments of the present invention.

The user inputs a piece of description text arbitrarily through the input module of the electronic device, and further selects the target country for which the trademark recommendation is to be generated. The system first determines whether the language of the description text is the same as the official language of the target country. If it is different, it will first modify the description text. Translate, and if they are the same, proceed directly to the follow-up process, and then analyze and extract the description text to produce keywords and industry-related text. The industry-related text is industry information.

The identification and extraction of the keyword part is mainly done by comparing it with the non-recognizable text database to confirm whether the keyword contains non-recognizable text. If so, the non-recognizable text will be removed and only Retaining identifiable text, specifically, the non-recognizing text database can be established by, but is not limited to, using a large language model database for training, and identifying commonly used words, particles, prepositions, etc. as non-recognizable.

On the other hand, the industry information (industry-related text) generated after analysis is used to generate an industry category corresponding to the description text input by the user through a trained language generation model.

The category recommendation module is used to compare trademark categories and product items based on the above-generated industry categories. Specifically, the comparison database will be different depending on the target country selected by the user. For example, the user selects the United States as the target country. , it will be compared with the official trademark category database of the United States. If Taiwan is selected, it will be compared with the official trademark category database of Taiwan. The purpose is because the product items in the trademark categories of different countries will be different. Sometimes it may happen that some product items in Taiwan cannot be found in the same category in the United States. Therefore, it is important to compare the trademark category database of the target country so that the recommended product items that are generated do not actually exist in that country. , this system will generate a comparison result of at least one trademark category recommendation information including at least one recommended category and at least one recommended product item based on the use of adjustable parameters and, for example, the similarity comparison of word vectors, to find similar trademark categories and product items. , to achieve the effect of accurate recommendation. The adjustable parameters are customized according to the user. The adjustable parameters can be selected from the options of user preferences, user habits, and the user's ability to tolerate risks.

The comparison results are then sorted by the report generation unit and filled in a modular format, thereby generating a trademark category recommendation report. Among them, the risk assessment module is used to compare the keywords extracted from the string with the trademark database of the target country and produce a risk information. This category recommendation module also includes a knowledge map unit. It is a dynamic learning model used to establish node matrix associations and evolve over time based on the equivalence relationship between product items in various countries and trademark items in various countries. The knowledge graph unit conversion at least includes the following steps: Convert multi-country trademark product items to Trademark project relationship data collection and dynamic model learning, through graph query and natural and grinding line reasoning, calculate the best match, shortest path, maximum flow, and optimize the equivalence relationship of product items based on the final decision results.

The report can be presented in the form of an online image, a downloadable document file, a file format that can be forwarded to communication software, a file format that can be shared to social software, an electronic bulletin board, or a combination thereof.

In Figure 24, the difference from Figure 23 is that after generating the trademark category recommendation information of the target country, the risk assessment module further performs a comparison of previous cases in the trademark category recommendation information through the search comparison unit. The trademark data of the target country Search and compare previous trademarks with similar identifying words in the database. The search and comparison unit is to find whether there are previously filed trademarks in the recommended category that are similar or similar to the identifying words extracted from the keywords. cases, and integrate and analyze the search results with recommended categories and recommended product items. For example, when the recommended category is Category 9, and the search comparison finds that Category 9 has similar previous cases, then the search comparison unit will compare and analyze the recommendations in detail. If there is no overlap between the product items and the product items in the previous case, the degree of similarity will be lower than if there is some overlap between the product items. The similarity calculation and arrangement of the search results and the identifying words are performed, and the similarity between the previous application category in the search results and the recommended information of at least one trademark category is calculated to generate an evaluation result.

Finally, the integrated analysis results are also put into a modular format through the report generation unit to generate a risk assessment report.

Furthermore, this system also establishes a well-known trademark database. In the above comparison, even if the previous case compared by the user and the search is in a different trademark category, if the previous case belongs to one of the well-known trademark databases, 1, the search and comparison results will still show the previous case, and will be prioritized and displayed as high risk.

Furthermore, after the category recommendation module parses the description text and extracts keywords and industry-related text to form industry information, it can also use the information acquisition unit to actively search external network information based on the extracted industry-related text. For companies with similar industry-related text descriptions, the company information resulting from the search and comparison is sent to a knowledge graph comparison unit of the category recommendation module, and then the trademark category comparison is performed through the knowledge graph comparison unit, and a generated At least one trademark category recommendation information, and the trademark category recommendation information at least includes: one recommended category and at least one recommended product item.

Please continue to refer to Figures 25 to 27, which are schematic diagrams showing usage scenarios of the system of the present invention.

As shown in Figure 25, in this usage scenario, the user plays the role of a general merchant, brand owner or trademark applicant. Through electronic devices such as mobile phones and computers, he connects to the Internet and activates the server through the web page. Enter any string describing the text in the field. These strings are semantically analyzed through the server in the cloud part. Specifically, it is first determined whether the language of the string and the official language of the target country are the same. If they are different, multilingual translation is used. The model is translated, and then performs string parsing to extract keywords and generate industry information, and then compares it with the trademark classification database of the target country through the category item comparison unit, and generates a trademark category recommendation report to feed back to the user vehicle.

Furthermore, the previous trademark cases are searched from the recommended trademark categories to find out whether there are previous trademark cases in the recommended categories and the similarity between the trademark and the product items. The search results are then combined with the recommended categories and recommended product items to generate risk assessment information. Finally, a risk assessment report is generated and fed back to the user's vehicle.

Furthermore, the trademark category recommendation report may be in the form of an online image, a downloadable document file, a file format that can be forwarded to communication software, a file format that can be shared to social software, an electronic bulletin board, or a combination thereof.

You can also crawl information from local servers such as official trademark application databases of various countries or any online information based on the input string, search for trademark records applied by other companies in similar industries, and obtain trademark categories and products. Items are also included as recommended options.

End users can provide this report to a professional firm or directly compile and submit the application documents to the official agency for trademark application.

As shown in Figure 26, in this usage scenario, the user's role is a professional trademark practitioner (Trademark Practitioners) or a firm (Trademark or law firm). The user connects to the Internet through a vehicle, usually an electronic device such as a computer. After that, activate the server and enter any descriptive text string through the field of the web page, and obtain the trademark category recommendation report or even the risk assessment report through the system of the present invention. The user can organize the report directly or again through the modular fixed format. A trademark proposal is then generated, and after automatically applying the format of the trademark practitioner or the law firm, it will be provided to the user's client. The client here is the brand owner, merchant or trademark applicant, and it can be confirmed by the client. Submit the application documents to the official agency to complete the case processing.

The schematic diagram of the risk assessment report can be found in Figure 27. It should be noted that Figure 27 is only an example of a usage scenario and does not limit the display method and format of the report. As shown in Figure 27, in the risk assessment report, similar trademark cases in the recommended categories will be listed, as well as their basic information such as trademark name/picture, country, filing date, applicant, classification and product items, official number, risk level, etc., through the display interface, the user can clearly know the risk level of the trademark to be applied for in the corresponding trademark category. Furthermore, the risk assessment for different target countries can be displayed by switching the target country. According to the report, the recommended categories in different target countries will most likely be the same, but the product items will be slightly different, and the similar trademark cases will also be different. Through the semantic analysis, automatic translation and judgment functions of the system of the present invention, it can Quickly generate multiple reports on different target countries, significantly reducing time costs and past category analysis and search thresholds.

It can be seen from Figure 25 and Figure 26 that whether the user is an end customer or an intermediate practitioner, the time cost of searching and trademark recommendation classification can be greatly shortened, and through the features of the system of the present invention: input any description text That is, you can use the multilingual translation model and text parsing unit to perform semantic analysis and text generation, understand the description text in a short time, and compare and approximate categories to generate recommended categories and product items, especially when applying for a cross-border application. play a greater role.

Finally, the technical features of the present invention and its achievable technical effects are summarized as follows:

First, through the semantic analysis trademark category recommendation system of the present invention, it can solve the problem that when ordinary users apply for a trademark online, they do not know which trademark category the goods sold or services provided belong to, and they do not know how to apply for a trademark for themselves. Select appropriate category items for the trademark, and provide users with quick choices through the trademark category recommendation report of the present invention.

Secondly, the semantic analysis trademark category recommendation system of the present invention solves the problem that users do not know whether there is a similar previous case for their trademark before applying for a trademark. The risk assessment report of the present invention provides users with immediate and easy-to-understand information. Understand the search results in an understandable way.

Third, through the comprehensive report generated by the semantic analysis trademark category recommendation system of the present invention, combined with trademark category recommendation and prior case retrieval, it is clearly marked in which category items have more or less trademark prior cases, and the colors can be used. Clearly distinguish or present it in any other markable way to provide users with clearer suggestions for selecting trademark category items.

It must be emphasized that the above detailed description is a specific description of possible embodiments of the present invention. However, the embodiments are not intended to limit the patent scope of the present invention. Any equivalent implementation or modification that does not deviate from the technical spirit of the present invention will All should be included in the patent scope of this case.

100: Electronic devices 110: Processor 120:Memory 130:Network interface controller 200:Internet 300:Server 310:Application 311:Trademark online application module 3111: Application data collection unit 3112:Category item selection unit 3113: Data transmission unit 312:Category recommendation module 3121: Text analysis unit 3122: Category item comparison unit 3123: Report generation unit 3124: Knowledge graph comparison unit 313: Data sending and receiving module 314:Risk Assessment Module 3142: Search comparison unit 400:Database 500:Sending server 600:Input module 601: Information retrieval unit 700: Multilingual Translation Model U:User S101, S102, S103, S104, S105, S106, S107, S108: steps S201, S202, S203, S204, S205, S206, S207, S208: steps S301, S302, S303, S304, S305, S306, S307, S308, S309: steps S401, S402, S403, S404, S405, S406, S407, S408: steps S501, S502, S503, S504, S505, S506, S507, S508, S509: steps S601, S602, S603, S604, S605, S606, S607, S608, S609, S610, S611: Steps S701, S702, S703, S704, S705, S706, S707, S708, S709: Steps S801, S802, S803, S804, S805, S806, S807, S808, S809, S810, S811, S812, S813: Steps S901, S902, S903, S904, S905, S906, S907, S908, S909, S910, S911, S912, S913; S914: Step

Figure 1 is a schematic diagram of the prior art; Figure 2 is another schematic diagram of the prior art; Figure 3 is another schematic diagram of the prior art; Figure 4 is another schematic diagram of the prior art; Figure 5 is another schematic diagram of the prior art; Figure 6 is another schematic diagram of the prior art; Figure 7 is another schematic diagram of the prior art; Figure 8 is a schematic diagram showing the system of the present invention; Figure 9 is a schematic diagram showing the electronic device of the system of the present invention; Figure 10 is a schematic diagram showing the server of the system of the present invention; Figure 11 is a schematic diagram showing the application program of the system of the present invention; Figure 12 is a schematic diagram showing the trademark online application module of the system of the present invention; Figure 13 is a schematic diagram showing the category recommendation module of the system of the present invention; Figure 14 is another schematic diagram showing the application program of the system of the present invention; Figure 15 is a schematic diagram showing the risk assessment module of the system of the present invention; Figure 16 is another schematic diagram showing the system of the present invention; Figure 17 is another schematic diagram showing the system of the present invention; Figure 18 is a flow chart showing the method of the present invention; Figure 19 is a flow chart showing another embodiment of the method of the present invention; Figure 20 is a flow chart showing another embodiment of the method of the present invention; Figure 21 is a flow chart showing another embodiment of the method of the present invention; Figure 22 is a flow chart showing another embodiment of the method of the present invention; Figure 23 is a flow chart showing another embodiment of the method of the present invention; Figure 24 is a flow chart showing another embodiment of the method of the present invention; Figure 25 is a flow chart showing another embodiment of the method of the present invention; Figure 26 is a flow chart showing another embodiment of the method of the present invention; Figure 27 is a flow chart showing another embodiment of the method of the present invention; Figure 28A is a schematic diagram showing an application example of the method of the present invention; Figure 28B is a schematic diagram showing an application example of the method of the present invention; Figure 29 is a flow chart showing another embodiment of the method of the present invention; Figure 30 is a flow chart showing another embodiment of the method of the present invention; Figure 31 is a schematic diagram showing the usage scenario of the method of the present invention; Figure 32 is a schematic diagram showing the usage scenario of the method of the present invention; and Figure 33 is a schematic diagram showing a usage scenario of the method of the present invention.

100: Electronic devices

200:Internet

300:Server

400:Database

500:Sending server

U:User

Claims (19)

A semantic analysis trademark category recommendation system for receiving a string provided by a user by operating an electronic device and at least one target country to be queried, through a processor of the electronic device through a network interface controller Connect to a server and execute an application to calculate and generate a trademark category recommendation report. The system at least includes: An input module is used to receive the string input by the user, label the string, and send a string of information; A category recommendation module is a computational model trained through natural language models and trademark classification tables and details to parse string information into trademark category recommendation information, which also includes: A text parsing unit used to parse the translated string information, extract keywords and industry information, and send them; A category item comparison unit is used to receive the keywords and industry information, compare it with the trademark category database of the target country, and calculate to generate at least one recommended category and at least one recommended product item; and A report generating unit is used to receive the recommended category, the recommended product item and the target country, and use the input language to generate a trademark category recommendation report through a templated format. A semantic analysis trademark category recommendation system as described in claim 1, wherein the system also includes a multilingual translation model, which is a computing model trained through large-scale language translation to receive the string information and determine the user input Whether the string is the same as the official language of the target country. If it is different, the string will be translated into the official language of the target country and all the translated string information will be sent. For example, in the semantic analysis trademark category recommendation system described in request item 1, the string may further include trademark text, trademark description, business behavior, company name, stock code, product name, service name, or a combination thereof. The semantic analysis trademark category recommendation system as described in claim 1, wherein the way the user inputs the string may be text input, voice input or video input. The semantic analysis trademark category recommendation system described in claim 1, wherein the input module further includes an information retrieval unit for actively retrieving relevant information from the tagged string information to the cyberspace, and This information can be sent to the multilingual translation model for translation. In the semantic analysis trademark category recommendation system described in claim 1, the trademark category database described in the category item comparison unit further includes the trademark category database of the target country and the international Nice trademark category database. The semantic analysis trademark category recommendation system as described in claim 1, wherein the system further includes a risk assessment module for comparing keywords extracted from the string with the trademark database of the target country, And produce a risk information. For the semantic analysis trademark category recommendation system described in claim 6, the risk information can be further integrated into the trademark category recommendation report through the report generation unit. The semantic analysis trademark category recommendation system as described in request item 1, wherein the report can be presented in an online image, a downloadable document file, a file format that can be forwarded to communication software, or a file format that can be shared to social software. , electronic bulletin board or combination thereof. For the semantic analysis trademark category recommendation system described in claim 1, the user can be selected from any combination of brand owners, trademark applicants, trademark practitioners, and legal practitioners. The semantic analysis trademark category recommendation system as described in claim 1, wherein the category recommendation module further includes a knowledge graph unit, which is used to establish a node matrix association, and according to the equality between product items and trademark items in each country. A dynamic learning model in which price relationships evolve over time. The knowledge graph unit conversion at least includes the following steps: collecting and learning dynamic model data on the relationship between trademark product items and trademark projects in multiple countries, and deducing the final result through graph query and natural and grinding reasoning. The best match, shortest path, maximum flow, and optimize the equivalence relationship of product items based on the final decision results. A semantic analysis trademark category recommendation method. The user operates an electronic device to query at least one target country, connects to a server through a network interface controller through a processor of the electronic device, and executes an application program to calculate The output method includes at least the following steps: (1) The user inputs a piece of description text arbitrarily through an input module of the electronic device; (2) A category recommendation module analyzes the description text and extracts keywords and industry-related text to form industry information; (3) Compare the parsed industry-related text with the target country’s trademark category database through a category recommendation module, output the comparison result based on an adjustable parameter, and generate At least one trademark category recommendation information of at least one recommended category and at least one recommended product item; and (4) using a report generation unit to generate a trademark category recommendation report from the trademark category recommendation information of step (3) in a templated format. The semantic analysis trademark category recommendation method described in claim 11, wherein step (2) may further include identifying text extraction of the extracted keywords through a text analysis unit , which can be divided into: steps (21) The text parsing unit generates and extracts identifiable text by comparing it with the non-identifying text database; and step (22) the text parsing unit connects the language model of the category item comparison unit to compare industry-related text Convert to generate industry categories. The semantic analysis trademark category recommendation method as described in claim 11, wherein step (1) further includes: Step (11) The user selects at least one target country for the trademark category recommendation to be generated; And further includes in step (3): Step (31) The category recommendation module compares the categories in the trademark category database of the target country based on the country selected by the user. The semantic analysis trademark category recommendation method as described in request item 11, which further includes after step (3): Step (301) uses a risk assessment module to perform a risk assessment on the identifying words, and uses a search and comparison unit to compare the identifying words with the target country in at least one trademark category recommendation information generated in step (3). The trademark database is compared to generate search results; Step (302) The report generation unit integrates and analyzes the recommended information of at least one trademark category and the search results, performs similarity calculation and arrangement on the search results and the identifying words, and simultaneously calculates the categories of previous applications in the search results and The similarity of the recommended information for at least one trademark category, thereby generating an evaluation result; and Step (303) The report generating unit combines the at least one recommended category and the assessment result to generate a risk assessment report through the templated format. The semantic analysis trademark category recommendation method as described in request item 12, which further includes after step (2): Step (21): Based on the industry-related text extracted in step (2), the information retrieval unit actively searches for and compares companies with similar industry-related text descriptions in external network information, and searches for the compared company information. A knowledge graph comparison unit sent to the recommendation module of the category; and Step (22) performs trademark category comparison through the knowledge graph comparison unit and generates at least one trademark category recommendation information. The semantic analysis trademark category recommendation method as described in request item 14, which after step (11) also includes: Step (12) The input module determines whether the input string language is the same as the official language of the target country. If not, it will first be translated through the multilingual translation model to translate the string language into the official language of the target country. Language The semantic analysis trademark category recommendation method described in claim 14, wherein the adjustable parameters are customized according to the user terminal, and the adjustable parameters can be determined by the user terminal preferences, user terminal habits, user terminal capabilities Choose from any of the options based on your ability to tolerate risk. The semantic analysis trademark category recommendation method described in claim 14, wherein the trademark category recommendation information at least includes: one recommended category and at least one recommended product item.

Images