Computer-implemented system for controlling workflows using complex text-based datasets

Abstract

The invention relates to a computer-implemented system comprising one or more computers, a database, and an input interface, wherein the computer-implemented system is designed to analyze the database and output, by means of a neural network using machine learning, proposals for carrying out new processes, the input interface makes it possible to input defined rules for individual processes, and the defined rules replace the proposals made by the neural network.

Description

[0001] Computer-implemented system for controlling workflows using complex text-based data sets

[0002] The invention relates to a computer-implemented system for controlling workflows using electronic data archives. In particular, the invention relates to such a computer-implemented system for controlling workflows in a patent law firm.

[0003] Data management systems to support the work of patent law firms, such as PATORG, GENESE, or ANAQUA, have been widely used in patent law firms for over 20 years. These systems are characterized by the fact that, in addition to bibliographic data on the various intellectual property rights, patents, trademarks, designs, or utility models handled by patent law firms, they also maintain an archive of correspondence with the respective files, both with offices and courts and with clients.

[0004] In recent years, due to the increasing prevalence of such systems, a trend has also developed among clients toward increasingly individualized and sophisticated communication. This concerns, among other things, the question of how documents should be exchanged—either in paper form, purely electronically via email, encrypted or unencrypted, via special electronic mailboxes, or directly through access to the database maintained by the patent law firm itself.

[0005] Likewise, a trend has developed, based on the practices of American law firms, for patent attorneys' services to be billed via electronic payment systems, whose predefined billing models harmonize only very partially with the fee schedules traditionally maintained by German patent law firms. Further complexity is introduced because some clients request their invoices in foreign currencies, such as US dollars.

[0006] Further special requests from clients, such as the parallel transmission of Word documents in addition to PDF documents or explicit instructions for invoicing, for example with each individual step, after completion of a more complex process or on a monthly basis, have led to a situation in which satisfactory interaction between law firms and their clients involves considerable effort and is largely dependent on the knowledge and skills of the patent attorneys involved.

[0007] The data management systems mentioned above attempt to address this problem by offering the option of implementing workflows that are designed to automatically process individual processes. A workflow is a predefined sequence of work steps triggered by an event.

[0008] For example, the receipt of an examination notice from the German Patent Office can trigger a workflow that asks the relevant employees to note and confirm the deadlines, starts the creation of a forwarding letter with reference to these deadlines and, if necessary, creates a forwarding invoice.

[0009] However, due to the specific client requests identified, workflows are only of limited use, as they can only manage complexity through a large number of queries. On the other hand, the number of possible workflows for a process—such as the receipt of an audit notice—increases immeasurably, and the respective employees must select the desired workflow for this client from the multitude of possible workflows. The object of the invention is therefore to provide a computer-implemented system that enables the reliable processing of such processes while using computer power in a resource-efficient manner.

[0010] The above-mentioned object is achieved by a computer-implemented system according to claim 1. The dependent claims relate to further advantageous aspects of the invention.

[0011] In the following, the invention is described using a preferred example with reference to the accompanying figures.

[0012] Figure 1 shows a flowchart illustrating the computer-implemented system of the invention; and Figure 2 shows a flowchart illustrating the operation of the system.

[0013] As shown in Figure 1, the system according to the invention comprises a computer network 1 with one or more terminals 3 (computers), one or more servers 5, and one or more databases 7. The system according to the invention can also be connected to an external server farm 9 or a cloud solution via a suitable network. Furthermore, suitable input interfaces or input devices, such as scanners 11 or Internet modems, can be connected to the system according to the invention.

[0014] At least one of the servers 5 is configured as a database server 15 for the data management system of the respective law firm.

[0015] The system further comprises an output device 13, via which the results can be displayed to a user. This can be a screen for text output or a loudspeaker for voice output. The system according to the invention is constructed in two steps.

[0016] Figure 2 shows schematically the individual steps for setting up the computer-implemented system.

[0017] In a first step, an analysis is carried out using the database of the law firm's data management system, taking into account both the content of the texts recorded there and the sequence of the respective work steps that were carried out in the respective file.

[0018] In a second step, a learned set of rules for work processes is then established based on the analyzed data.

[0019] A work step within the meaning of the invention is considered to be, for example, noting a deadline or creating a task.

[0020] A task within the meaning of this invention is understood to be a computer-generated message to a specific addressee, with which the addressee is requested to process a specific process.

[0021] In a third step, this learned set of rules is continuously trained through machine learning based on user feedback.

[0022] In a fourth step, a predefined set of rules is created based on the information made available to the public by the patent and trademark offices, such as the list of all possible official decisions and the rules for calculating time limits applicable to the respective official decisions. This predefined set of rules is incorporated in a final step and replaces the learned set of rules at certain points.

[0023] In other words, the system according to the invention first records the entire data stock in the law firm's data management system.

[0024] When a new process is entered, such as an official notice, the system examines how similar situations have been handled in the past and makes a corresponding suggestion by creating the necessary tasks to process the process.

[0025] For example, the German Patent and Trademark Office sends a bibliography notification after a patent application has been filed. This notification is captured by the system according to the invention via a suitable interface 11, either by electronic transmission (DPMA-PRO) by the office or by a scanner with subsequent OCR conversion in the office, and compared with the data stored in the data management system.

[0026] In the event of a deviation, a task is created for an employee to adjust the data either on the Patent Office side or in the firm's internal data management system.

[0027] If the data matches, one of the attorneys is assigned the task of signing a report letter, which can be provided as a draft based on the same processes in other files. Since there are no official deadlines associated with this process, an internal follow-up is also noted to monitor the forwarding of the bibliography notification with the associated report letter to the client. The system according to the invention proceeds similarly for the subsequent transmission of the publication of the application or if the official file number has already been sent earlier.

[0028] Until now, it was common practice to program these processes using workflows that were firmly linked to corresponding rules.

[0029] However, for the reasons described above, this involves considerable effort and difficulties.

[0030] According to the invention, artificial intelligence is therefore used at this point, which receives the entire database or at least part of the database of the respective patent law firm as input and, on the basis of the processes depicted there, creates a proposal for processing the respective communication.

[0031] Through feedback from employees on these suggestions, the system can learn and thus ensure reliable handling of the respective messages in the shortest possible time.

[0032] In addition, the system can be given fixed rules, such as in the case of the bibliography notification described above, for comparing the data stored in the data processing system with the bibliography data transmitted by the office. This is useful when a specific work step must always be carried out in connection with a process.

[0033] However, the further processing of the report letter - be it as a paper letter, a PDF document, an email, encrypted or not -, the transfer to the client's electronic mailbox or making the document available to the client for downloading, is set up using machine learning. This is useful when a number of different work steps are possible. In other words, based on the details of a particular file, such as the name of the intellectual property rights holder, the name of any foreign correspondent law firm that may be involved, the relevant filing office, etc., the system quickly learns from comparison files whether a letter should be generated in the relevant case, an email is sufficient, or a notice should be sent to the client that a corresponding document is available for download.

[0034] This learning also incorporates the content of client instructions, which they have sent to the firm in the form of letters or other messages. To this end, the inventive computer-implemented system analyzes the electronic archive of the data management system, recognizes the client's instruction letters, extracts the respective instructions, and, taking into account the chronological order of these instructions, adapts the workflows.

[0035] The system according to the invention can, on the one hand, enable the specification of rules, and, on the other hand, determine the handling of individual processes by means of a neural network based on the law firm's archive and machine learning, so that overall computer resources are saved, compared to the case where the entire system would be based only on a neural network without the possibility of specifying fixed rules.

[0036] On the other hand, the work required to implement the system can be simplified considerably compared to the case where a complete set of rules has to be created manually.

[0037] Furthermore, the system is configured to specifically analyze the content of client letters to the law firm for specific instructions and incorporate them into workflows. For example, the system is capable of recognizing a client's instruction requesting that they no longer receive paper documents and integrating them into the workflow for that client accordingly. These instructions can then override the previous procedure established through machine learning.

[0038] In more complex cases, such as the receipt of an examination decision from the Patent Office, the system can determine the deadline for responding to the decision based on the date the examination decision was received in the law firm's electronic mailbox and on the analysis of the text of the official decision and store it as an initial entry in the law firm's deadline calendar.

[0039] To ensure security, a task is always generated for the employees of the office in order to check and confirm the deadlines using the four-eyes principle.

[0040] At the same time, in the case of a non-German-speaking client, a task is created to prepare a translation of the Zimtsbescheid for an employee of the firm.

[0041] For control purposes, an internal forwarding date for the translation is also noted in the deadline calendar.

[0042] The system will then generate a report letter and forward it to one of the attorneys as a task for signature. At the same time, the attorney handling the case will be assigned a further task to draft a commentary for the client, depending on whether the client requests such a commentary. The system will determine this based on comparative files from the past and on the client's letters of instruction.

[0043] In practice, the lawyer can be informed whether the client would like to have a draft response to the notice drawn up straight away, whether a short comment is sufficient or whether the client wishes to forgo a comment altogether.

[0044] Together with the report letter, the system will create a corresponding cost invoice based on the costs entered by the processors in the file's cost sheet and on the firm's fee schedule - for example, for noting deadlines and forwarding documents - if the client requests an invoice for each individual step.

[0045] If the client only wishes to be billed once a month or only after an official notice has been dealt with, the costs will remain recorded in the cost sheet to be billed later.

[0046] The system according to the invention can also be used to analyse incoming mail from clients or foreign colleagues who represent corresponding intellectual property rights before foreign patent offices on behalf of the law firm.

[0047] Incoming letters or emails are "read" by the system and checked for deadlines. For example, if a client requests a draft for a reply to a legal notice, which is to be delivered to them by a specific time, the computer-implemented system of the invention will note the deadline requested by the client in the firm's deadline calendar and generate a task for the attorney responsible for the file.

[0048] At the same time, the system can send a confirmation letter to the client confirming receipt of the instructions and indicating the deadline for the draft.

[0049] The system can also perform an initial conflict check for new assignments to determine whether a conflict of interest exists in the performance of a mandate. If this is the case, a corresponding warning message is sent to the firm's attorney(s).

[0050] The typical mixture of standardized procedures for patent law firms, which result from the specifications of the patent offices and the highly standardized decisions issued there, on the one hand, and the high degree of flexibility expected by clients, can be processed safely and reliably through a combination of predetermined rules and a neural network based on AI application, on the other hand.

[0051] Although the invention has been described above using the example of a patent law firm, the invention is not limited to this. Similar problems also arise in tax offices or law firms and can be solved using the means described.

Claims

Patent claims 1. Computer-implemented system comprising: one or more computers (3), a database (7), an input interface (11), and an output device (13), wherein the computer-implemented system is designed to analyze the database (7) and, with the aid of a neural network, to output suggestions for processing new processes by means of machine learning, the input interface (11) enables the input of fixed rules for individual processes, and wherein the fixed rules replace the suggestions developed by the neural network.

2. The computer-implemented system according to claim 1, wherein the database (7) comprises the archive of a file management program for a patent attorney's office, tax office or law firm.

3. The computer-implemented system according to claim 1 or 2, wherein data entered via the input interface (11) is analyzed for deadlines, and these deadlines are noted in a deadline calendar according to their importance.

4. The computer-implemented system according to any one of the preceding claims, wherein the system is configured to automatically issue tasks to respective users.

5. The computer-implemented system according to claim 4, wherein the system is configured to check the consistency of the input data with data already recorded in the system. review and, where appropriate, issue a task to align the data.

Images