Procedure for evaluating official examination certificates regarding industrial property rights and computer program for this purpose

Abstract

The invention relates to a method for evaluating official examination certificates concerning industrial property rights, comprising the following steps: providing the official examination certificate as an electronic file, text analysis of the official examination certificate, AI-supported evaluation of the results of the text analysis to identify and assess listed obstacles to the granting of protection by a plurality of dedicated AI agents, calculation of respective scores based on the identified and assessed obstacles to the granting of protection, database analysis using a historical database of other industrial property rights to classify the examined industrial property right with regard to further relevant parameters.Calculation of the expected workload for the successful completion of a protection granting procedure based on the score and other relevant parameters, and output of this expected workload and AI-supported evaluation.

Description

The invention relates to a method for evaluating official examination certificates concerning industrial property rights and an associated computer program. When applying for and examining industrial property rights, such as patents, it is customary for the responsible authority or examiner to issue an official decision at the beginning of the examination process. This decision may take the form of an opinion-based search result or an initial examination report. This report summarizes and explains the main obstacles to granting protection. The official examination report is then analyzed by a patent attorney or the applicant's in-house specialists to determine whether to pursue the patent application further, considering the anticipated costs of the examination process and the chances of success, or whether to abandon the patent application at an early stage. When a person makes a decision, a number of personal factors and attitudes dependent on their state of mind on any given day play a role, meaning that the result is not reliably reproducible. Furthermore, it is difficult for a person to grasp and weigh the many factors involved in examining an industrial property right. The invention therefore aims to provide a tool that enables the appointed patent attorney or in-house specialist staff to objectify and reproducibly make the decision. This problem is solved by a method according to claim 1 and a computer program according to claim 9. The dependent claims relate to further advantageous embodiments of the invention. In the inventive method for evaluating official examination certificates relating to industrial property rights, the official examination certificate is provided as an electronic file, for example as a PDF or Word file; the official examination certificate is text-analyzed; the result of the text analysis is evaluated using AI to identify and assess the listed obstacles to granting protection by a plurality of dedicated AI agents, whereby a "Large Language Model" can be used; a score is calculated based on the identified and assessed obstacles to granting protection; and a database analysis is performed using a historical database of other industrial property rights to classify the industrial property right under investigation with regard to further relevant parameters that influenced the examination procedure in previous methods.The calculation of the expected workload for the successful completion of a protection grant procedure based on the score and other relevant parameters, whereby the historical database with data on previous procedures can also be accessed, and finally an output of this expected workload and the AI-supported evaluation. Preferably, the other relevant parameters include one or more of the following groups: the scope of the application text of the industrial property right, the number of pending claims, the number of prior art documents listed in the official examination report and / or their scope, an AI-determined quality assessment of the application text based on historical comparisons of industrial property rights by the same author, an AI-determined examination performance of an assigned official examiner and / or examination division based on historical comparisons, an AI-determined assessment of protectability based on examination results of other family members of the industrial property right, and an AI-determined weighting factor for an IPC class assigned to the industrial property right. In the method according to the invention, the plurality of dedicated AI agents can include one or more of the following groups: a clarity agent that analyzes possible clarity objections in the official examination report, an eminentity agent that analyzes objections regarding the disclosure required for eminentity, a unity agent that analyzes objections regarding lack of unity of the claimed invention, a novelty agent that analyzes objections regarding lack of novelty, an inventive step agent that analyzes objections regarding lack of inventive activity or inventive step, a technicality agent that analyzes objections regarding lack of a technical teaching, and an exclusion agent that analyzes other objections regarding further exclusion criteria for patentability. To improve the interpretability of the output, it is possible to present the workload in the form of summed hourly rates of the specialists assigned to the task. In the method according to the invention, one or more of the AI ​​agents can be designed as a neural network. Alternatively or additionally, one or more of the AI ​​agents can use machine learning. To enable a quick interpretation of the output, it is preferred to display the result of the AI ​​agents in the form of a traffic light, i.e., in the form of respective colored indicators of the severity of the respective listed obstacle to granting protection. In a further development of the inventive method, a further AI agent is provided which verifies the validity of the cited obstacles to granting protection by comparing the application text and the disclosure of each of the prior art documents. Finally, the invention can be implemented as a computer program which, when loaded into the working memory of a computer, causes it to execute a method according to any one of claims 1 to 8. The invention is described below with reference to a preferred embodiment and the accompanying figures, in which: Fig. 1 shows a flowchart of the method according to claim 1, Fig. 2 shows an example of the inventive output of the method according to claim 1. The inventive method for evaluating official examination certificates relating to industrial property rights is set up as a computer program either on a local computer or in a cloud solution on an outsourced server. The procedure begins with the receipt of an official examination notice by the applicant or their representative of an industrial property right, which for the sake of simplicity will be referred to below as a patent application, but is not limited to this. Other industrial property rights, such as utility models, are also included, provided they are subject to examination. Preferably, the official examination certificate is issued by the authorities as an electronic document and sent to the applicant or their representative. Alternatively, an official examination certificate sent in paper form can also be scanned by the recipient and converted into an electronic document. This electronic document is made available for the procedure in a first step by copying it into an inbox of the computer program or by dragging and dropping it in the usual way. Subsequently, unless the official examination notice is encoded as a text document, the electronic document is subjected to character recognition using a known OCR step. Under the control of an initial AI agent, designated "CEO," the AI ​​analysis of the document then begins. The CEO calls upon a series of agents, either sequentially or concurrently, to examine the official examination report for the presence or absence of any specific obstacle to granting protection. For example, a second AI agent checks the official examination report to determine whether the issuing examination department or examiner has raised any clarity objections to the claims. The AI ​​agent can also assess the number of different clarity objections raised and estimate the effort required to address them. This estimate is based on training data previously provided to the agent, either through a database of other completed examinations or through feedback from a user training the agent.The agent can provide the result as a kind of "score" or point value. Another AI agent checks the official notification for any objections regarding a lack of uniformity. Here, too, the result is provided as a score. The effort required to address an objection based on a lack of uniformity typically involves rewriting the claim set, excluding at least one of the other inventions, and revising the description to delete or appropriately identify embodiments no longer covered by the claim set. This effort is also determined using training data, incorporating further parameters, such as the scope of the description, which must be revised with regard to embodiments no longer covered. Another AI agent examines the official examination notice regarding objections to the feasibility of the invention and determines a score in a similar manner. Other AI agents examine the official examination notice for objections regarding lack of novelty, lack of inventive step, and other grounds that could prevent the granting of the intellectual property right, and calculate respective scores based on their training data. The official examination notice is therefore analyzed sequentially or in parallel by, for example, a clarity agent, an enablement agent, a unity agent, a novelty agent, an inventiveness agent, a technicality agent, and an exclusion agent. According to the invention, the computer program has access to a database, such as the file management program of the patent law firm representing the intellectual property right, and an AI database agent can search this database for similar cases, i.e., cases with a similar distribution of scores. Furthermore, the cases in the database used for comparison can be restricted based on additional parameters, for example, by considering only cases of the same IPC class or only cases where the application was written by the same person skilled in the art. Since the AI ​​database agents are implemented as neural networks in a preferred embodiment, they can determine further parameters from the database that influence the result and use these to select the cases to be considered.These parameters include, but are not limited to, the scope of the application text of the industrial property right, the number of pending claims, the number of prior art documents listed in the official examination report and / or their scope, an AI-determined quality assessment of the application text based on historical comparisons of industrial property rights by the same author, an AI-determined examination performance of an assigned official examiner and / or examination division based on historical comparisons, an AI-determined assessment of protectability based on examination results of other family members of the industrial property right, and an AI-determined weighting factor for an IPC class assigned to the industrial property right. For example, an AI agent can determine the scope, i.e., the number of pages, of the prior art references cited in the official examination report and estimate the effort required by the patent attorney to study these references. In the case of a particularly preferred embodiment, an AI assistant will also examine the prior art references, or at least the references cited in the official examination report, to determine whether they justify the objection, since the effort required to refute an objection through argumentation is generally less than the effort required to address it by amending the claims. Based on the scores of the individual agents, a relative weighting of the individual scores learned through training, and the historical database, the computer program determines the workload in terms of working hours and outputs this result as the sum of hourly rates for the specialists assigned to the task. The program can take into account that different specialists may have different hourly rates, for example, a patent draftsman when corrections to the drawings need to be made, or a translator when, for instance, the translation of a priority document is required. In operation, the computer program will be preferred and will perform machine learning, at least in part, by comparing its forecasts with the actual effort incurred up to the point of issuance and using this information to improve the individual AI agents and the CEO agents. To enable the user to quickly grasp the analysis results, it is preferable to convert the scores of the individual AI agents into a color code and display them, for example, as a traffic light with three levels. Green indicates no issues within that AI agent's area of ​​responsibility, red indicates very serious issues, and yellow indicates issues requiring moderate effort. The color system is not limited to this, and more levels or even continuous output models can be used, as shown in Figure 2. Figure 2 shows, for example, the evaluation of the inventive method in the case where no objections were raised regarding clarity, feasibility, technicality, novelty, or inventive step, but an objection was raised due to a lack of unity. Based on its training data, the scope of the description, and the number of affected patent claims, the AI ​​agent for unity outputs a yellow value, which corresponds to a low average effort. Using this training data, the CEO determines that all objections can be resolved in a single step, i.e., with a single reply, and estimates the time required for this reply and the associated hourly rates, for example, at 1500 S. Although the invention has been described using a preferred embodiment, it is not limited to this embodiment, and further modifications and adaptations are possible. For example, the method can also include access to public patent office databases to consider the examination results of parallel intellectual property rights in other countries. Access to patent office databases is also conceivable to obtain information about the examination procedures of the respective examiner or examination department, such as whether minor formal objections are resolved independently by the examining authority or whether new official examination reports are issued in each case.The examiner's or examination department's inclination to invite the applicant to a hearing, and whether a video conference or a personal appearance is preferred, can also be determined in this way if the patent law firm's database cannot itself provide enough comparable cases.

Claims

Procedure for evaluating official examination notices regarding industrial property rights, comprising: providing the official examination notice as an electronic file, text analysis of the official examination notice, AI-supported evaluation of the text analysis results to identify and assess listed obstacles to granting protection using a plurality of dedicated AI agents, calculation of respective scores based on the identified and assessed obstacles to granting protection, database analysis using a historical database of other industrial property rights to classify the examined industrial property right with regard to further relevant parameters, calculation of the expected workload for the successful completion of a protection grant procedure based on the score and the other relevant parameters, and output of this expected workload and the AI-supported evaluation. The method for evaluating official examination reports concerning industrial property rights according to claim 1, wherein the further relevant parameters include one or more of the following groups: the scope of an application text of the industrial property right, a number of pending claims, a number of prior art documents listed in the official examination report and / or their scope, an AI-determined quality rating of the application text based on historical comparisons of industrial property rights by the same author, an AI-determined examination performance of an assigned official examiner and / or examination division based on historical comparisons, an AI-determined assessment of protectability based on examination results of other family members of the industrial property right, and an AI-determined weighting factor for an IPC class assigned to the industrial property right. The method for evaluating official examination decisions relating to industrial property rights according to claim 1 or 2, wherein the plurality of dedicated AI agents include one or more of the following groups: a clarity agent, an executability agent, a unity agent, a novelty agent, an inventiveness agent, a technicality agent and an exclusion agent. The procedure for evaluating official examination notices concerning industrial technical property rights according to one of the preceding claims, wherein the output of the work effort is in the form of summed hourly fees of the specialists designated for processing. The method for evaluating official examination notices regarding industrial technical property rights according to one of the preceding claims, wherein one or more of the AI ​​agents are designed as neural networks. The method for evaluating official examination notices regarding industrial technical property rights according to one of the preceding claims, wherein one or more of the AI ​​agents use machine learning. The procedure for evaluating official examination notices regarding industrial technical property rights according to one of the preceding claims, wherein the output of the AI-supported assessment is in the form of respective colored displays of the severity of the respective listed obstacle to granting protection. The method for evaluating official examination decisions regarding industrial property rights according to one of the preceding claims, wherein a further AI agent verifies the validity of the cited obstacles to granting protection by comparing the application text and a disclosure of each of the prior art documents. A computer program which, when loaded into the working memory of a computer, causes the computer to execute a method according to any one of claims 1 to 8.

Images