Web application operation support system, method, and program

Abstract

This provides a web application operation support system, method, and program that can continue to operate even after changes to the UI structure. [Solution] The method includes the steps of: obtaining the structure and operable elements of each page of the target web application; describing each page as structured metadata consisting of multiple categories including subject, operation type, target area, constraints, and context; extracting and associating semantically identifiable features for each operable element in addition to location information; saving the metadata in a referable format; and enabling continuous operation even after a change in the UI structure by having a processing means operating in the user's browser environment convert the user's natural language input into the same format as the structured metadata, and when identifying an operable element by referring to the metadata, searching for similar elements using semantically identifiable features if a search using location information fails.

Claims

1. A method for enabling natural language operation without modifying an existing web application executed by a computer, The computer obtains the structure and manipulable elements of each page of the target web application, and describes each page as structured metadata consisting of multiple categories including subject, operation type, target area, constraints, and context. The computer performs the steps of extracting and associating semantically identifiable features for each operable element, in addition to location information. The computer saves the metadata in a format that can be referenced, A processing means operating in the user's browser environment converts the user's natural language input into the same format as the structured metadata, and when identifying the target element by referring to the metadata, if a search using location information fails, it searches for similar elements using the semantically identifiable features, thereby enabling continuous operation even after a change in the UI structure. A method for supporting the operation of a web application, characterized by including the following.

2. In claim 1, The aforementioned multiple categories consist of five categories: Actor, Operation, Scope, Constraints, and Context. A method for supporting the operation of a web application, characterized by the features described herein.

3. In claim 2, The aforementioned acquisition step is, The computer analyzes the navigation area within the seed URL page and extracts the links within that area as priority targets. The computer includes the steps of: detecting action buttons on each page, detecting forms that are dynamically expanded when such buttons are clicked, and recording semantic features for each input field in the form, including label text, attribute information, and neighboring text. The computer extracts the help text, tooltips, and explanatory text contained in each page and stores them in association with the metadata. A method for supporting the operation of a web application, characterized by including the following.

4. In claim 3, The aforementioned acquisition step is, The computer, in response to a web application requiring authentication, temporarily receives authentication information at the start of the acquisition process, automatically identifies the login form, and completes the authentication. The computer, after completing authentication, discards the authentication information from memory and does not permanently store it; The computer then navigates through all pages while maintaining the same browser session. A method for supporting the operation of a web application, further comprising the following:

5. In the method according to claim 1, The computer then performs the steps of estimating the operations that can be performed on each page from the metadata and element information of each page obtained, The computer automatically generates test scenarios described in natural language for each of the estimated operations, The computer saves the generated test scenario in association with the metadata, The computer compares the existing test scenario with the newly acquired results during reacquisition and automatically updates the test scenario. It further includes, Automatically generates operational tests for the target web application in natural language format from acquired results, without the need for manual test case definition. A method for supporting the operation of a web application, characterized by the features described herein.

6. In claim 1, The computer stores and manages the metadata in a database on a tenant-by-tenant basis. The computer issues a script corresponding to the tenant and embeds it within the page of the target web application, thereby deploying an agent that operates within the user's browser. The script includes an identifier unique to the tenant, and the agent, upon startup, uses the identifier to retrieve the latest metadata corresponding to the tenant from the server, and if it detects a change in the UI of the target application, it performs a differential update by re-retrieving the data. A method for supporting the operation of a web application, further comprising the following:

7. In claim 1, The aforementioned semantically identifiable features include display text, element attribute information, relationships with neighboring elements, and relative position within the form. A method for supporting the operation of a web application, characterized by the features described herein.

8. In claim 1, The computer defines a control policy as an external parameter, which includes the risk classification of the operation, whether or not pre-execution verification is required, and the grace period for correction. The computer distributes and applies the control policy to the processing means, thereby reserving the user with final execution rights in update operations. The processing means, in applying the control policy, includes the steps of classifying the interpreted operations into risk levels, requiring explicit confirmation for high-risk operations, and assigning medium- and low-risk operations to be automatically executed after a correction grace period, The processing means includes the step of variably setting the length of the correction grace period based on the risk level of the operation and the reliability of the analysis, The processing means includes the steps of: resetting the grace period if additional input is detected during the correction grace period; canceling execution if an interruption instruction is detected; and automatically executing the operation if the grace period has elapsed without interruption; A method for supporting the operation of a web application, further comprising the following:

9. A self-healing operation support system, An input conversion unit that converts the user's natural language input into structured data, A metadata storage unit that holds metadata describing the page structure and operable elements of the target web application, The aforementioned operable element includes an element information holding unit that holds positional information and semantically identifiable features, A determination unit that determines the next action based on the structured data and metadata, If the search using the aforementioned location information fails, an element selection unit rediscovers the target element based on the semantically identifiable features, An execution unit that performs an operation on the target element identified by the element selection unit, A self-healing operation support system characterized by comprising the following features.

10. In claim 9, The aforementioned system is implemented as a script embedded within a page of the target web application. The metadata storage unit retrieves the latest metadata from the server based on the tenant identifier. When a UI change is detected in the target application, a differential update is requested from the server. A self-healing operation support system characterized by the following features.

11. In claim 9, The element selection unit is, Search by location information, Searching by identifier attribute assigned to an element, Searching for similar elements based on the aforementioned semantically identifiable features, The following will be executed in this fallback order: Similarity is calculated by weighted scoring of the aforementioned semantically identifiable features. If multiple candidates have similar scores, the system will switch to user confirmation or candidate presentation depending on the risk level of the operation. A self-healing operation support system characterized by the following features.

12. In claim 9, The aforementioned input conversion unit is As user utterance operation types, data manipulation (create, retrieve, update, delete) and explanation requests (explain) are analyzed using the same 5-slot structured data format. The determination unit, For data manipulation, the system outputs a result indicating whether to perform a DOM operation. In response to a request for explanation, the system searches for knowledge information associated with the metadata storage unit and outputs a determination result that provides the answer. The aforementioned system, Maintain conversation history in continuous dialogues that mix data manipulation and explanation requests, and enable switching between manipulation and explanation while maintaining context. A self-healing operation support system characterized by the following features.

13. In claim 9, An operation pattern database that stores patterns of previously successful operations as vector data, A means for searching the operation pattern database for new user input and, if a similar pattern exists with a confidence level above a predetermined threshold, for reusing the similar pattern without calling the language model, A means for performing operational analysis using a language model when the number of similar patterns is below a predetermined threshold, or when no suitable pattern exists. If the operation analysis by the language model is successful, means for saving the operation pattern based on the results of the operation analysis to a database, A means for recording the number of uses and successes of the aforementioned operation pattern and dynamically updating the reliability level, A self-healing operation support system characterized by comprising the following features.

14. In claim 9, The determination unit, If the target of the operation is not explicitly stated from the user's input, a means of inferring the target of the operation by referring to past conversation history or operation history is provided. Means for setting an identifier indicating the source of the inference for the inferred target of operation, When the aforementioned identifier indicates an inference from the conversation history and the operation type is a write operation, a means for confirming the inferred operation target with the user, Equipped with, If the user confirms, update the identifier to an explicit specification and execute the operation. For inputs where the target of the operation is explicitly specified, the operation will be performed immediately without confirmation. A self-healing operation support system characterized by the following features.

15. In claim 9, The determination process is executed based on an external parameter file that controls the operation of the determination unit. A means to analyze the cause of failure when the judgment result differs from the expected result, based on the execution results of the test cases. A means for automatically generating suggestions for adjusting the external parameters according to the failure pattern, The aforementioned adjustment proposal includes means to include an explanation of the side effects of the adjustment, Means for automatically reflecting the aforementioned adjustment proposal in the external parameter file based on user approval or when predetermined safety conditions are met, Furthermore, This allows for improvement of the judgment logic by adjusting external parameters only, without modifying the program itself. A self-healing operation support system characterized by the following features.

16. A method for automating the testing of a web application using the system described in claim 9, The computer converts test cases written in natural language into structured data using the input conversion unit, The computer, using the element selection unit, identifies the element to be operated on based on the semantically identifiable features rather than location information; The computer performs an operation on the identified element and records the test results. Includes, The computer enables the test to continue running without redefining the test cases, even after the UI of the target web application has been changed, by rediscovering elements based on the semantically identifiable features. A test automation method characterized by the following:

17. A program for causing a computer to function as a component of the self-healing operation support system described in claim 9.

18. In claim 9, The determination unit, The structured data derived from user input and the metadata of the current page are both stored as multiple slots corresponding to multiple categories including subject, operation type, target area, constraints, and context. For each of the aforementioned multiple slots, the difference is detected as a gap on a slot-by-slot basis by determining whether it matches, does not match, or is missing. A self-healing operation support system characterized by the following features.

19. In claim 18, The determination unit, The detected types or combinations of gaps are classified as gap patterns. The next action is determined based on the action type associated with the gap pattern. A self-healing operation support system characterized by the following features.

20. In claim 19, The correspondence between the gap pattern and the action type is stored in an external parameter file as a declaratively defined correspondence table, rather than as a conditional branching logic. By changing the external parameter file, the determination logic of the determination unit can be adjusted without changing the main program. A self-healing operation support system characterized by the following features.

Images