[022] As will be appreciated, a highlighted part of equation (1), i.e., (id=’35345’) and (2), i.e., (class=”nav-bar”), represents two different types of DOM elements that may be extracted from the application code component. In other words, equation (1) and (2) represents two different types of DOM elements extracted that may be used for developing RPA bots. Thereafter, the plurality of DOM elements may be activated, and a message may be passed to the element tracking module 206 for fetching the plurality of DOM elements. The element tracking module 206 may fetch the plurality of DOM elements by parsing the application code component for the repetitive pattern. Moreover, detection and tracking of the plurality of DOM elements between the browser application and the application code component may be performed based on auto-inspection of the plurality of DOM elements. The web UI extracting module 204 may activate the element tracking module 206 by sending the webpage details and the application code component.
[023] Thereafter, the element tracking module 206 may determine the plurality of DOM elements. Additionally, the web UI extracting module 204 may receive the plurality of DOM elements from the optimized testing interface 220 for learning purpose. The web UI extracting module 204 may communicate to the element tracking module 206 via several network protocols used on Internet, including Transmission Control Protocol/ Internet Protocol (TCP/IP), User Datagram Protocol (UDP), Simple Network Management Protocol (SNMP), Internet Control Message Protocol (ICMP), Secure Socket Layer (SSL) and Hypertext Transfer Protocol (HTTP).
[024] The element tracking module 206 may be configured to receive the web page details and the application code component from the web UI extracting module 204. Moreover, the element tracking module 206 may be configured to trace each of the plurality of DOM elements from a section of the application code component. The element tracking module 206, may trace each of the plurality of DOM elements by inspecting activated state of each of the plurality of DOM elements. In addition, the element tracking module 206, may be configured to consume an active session from the web UI module 204, that may be subjected to more complexity. The active sessions may be determined from the activated state of each of the plurality of DOM elements. It should be noted that, many a times due to different script and operations, the plurality of DOM elements may remain active that may be needed to be avoided by the element tracking module 206, in order to reduce capturing of a false positive sessions. In an embodiment, the element tracking module 206 may reduce the complexity caused due to consumption of the active sessions by implementing a thread for each of the active session. Further, the thread implemented may implement a counter for each of the active session. The counter implemented may not increase, when the active session may operate in its own cycle. However, the counter may increase, when the active session may pass from one active session to another active session, thereby mitigating the thread to that active session. This in turn may remove the false positive sessions from the active sessions.
[025] Moreover, the element tracking module 206 may act as an extension for tracking respective element from the application code component. In order to track the respective element, the element tracking module 206 may match the respective element with the associated application code component. When the element and the associated code component is matched then, a token may be generated correspond to the application code component. Thereafter, the element tracking module 206 may send the respective token to the parsing module 210 in order to parse the element from the associated code component. Moreover, when the element does not match with the associated code component, then the application code component may be checked. However, sometimes due to dynamic state of the element, the element may not match exactly but rest part of the application code component may match. This may help to track location of the element in the application code component. The tracked location of the element may be remembered by implementing a pointer. Moreover, the element tracking module 206 may be configured to send the plurality of DOM elements traced, which may be derived from the application component for storage to the data repository 208.
[026] The data repository 208 may be configured to receive the plurality of DOM elements traced that may be derived from the application code component from the element tracking module 206. In an embodiment, the data repository 208 may act as a temporary storage unit. The data repository 208 may store a dataset and may be responsible to perform all sorts of parsing and extraction operations. Moreover, the data repository 208 may perform state detection and classification by initializing temporary memory. The data repository 208 may operate on instances. By way of an example, there might be certain cases where multiple web-drivers may be needed to be operated at a same instance. In such cases, the data repository 208 may include temporary memory that may be divide based on instances of the web-drivers along with the domain with which the web-drivers may be connected. Moreover, a new memory in the data repository 208 may be allocated for every new web driver instanced along with a new domain. However, the web-drivers that may connect to a similar domain even with different functionalities may be kept in same instance of the temporary memory. This may help in increasing performance by reducing repetitive task of initialization of defining the application code components as the initialization for the repetitive task may be already available.
[027] The parsing module 210 may receive the application code component of the repetitive pattern and the plurality of DOM elements traced by using an application driver / the web-drivers from the data repository 208. Once the application code component may be received, the parsing module 210 may remove a code from the dataset so as to fetch an output code. The output code may be extracted by determining a source for each of the plurality of DOM elements. It should be noted that, the source for each of the plurality of DOM elements may be defined in a cloud environment by the web-drivers and the active sessions. Moreover, the web- drivers share the source of the plurality of DOM elements, when computation for each of the plurality of DOM elements may be performed locally.
[028] However, since different browser applications may have different mechanisms for initializing the application code component, hence, the parsing module 210 may determine change in each of the different mechanism based on a category of the browser application. By way of an example, in case of the chrome application, the web-drivers of the chrome application may use different forms of initialization of the application code component. Similarly, for the fire-fox application, the web-drivers of the fire-fox application may use different forms of initialization of the application code component. It should be noted that, initialization of the application code component may vary from one browser application to another. Additionally, another parameter that must be considered for having different mechanisms for initializing the application code component may be related to hosting tokens. Another parameter considered related to the hosting tokens may be due a personalized hosting server i.e., there may be different mechanisms for initializing the application code component. As will be appreciated, in order to avoid different mechanisms that may be used for initializing of the application code component for different browsers, the end user may need to install web-drivers while using the parsing module 210. Further, the parsing module 210 may use the web-drivers to determine the hosting tokens for the corresponding browser application. By way of an example, based on the web-drivers used to determine the hosting tokens, the parsing module 210, may extract the output code based on a pre-defined rule of Hypertext Markup Language (HTML) and eXtensible Markup Language (XML). It should be noted that, the pre-defined of HTML and XML may be used to dissolve the application code component from the data repository 208.
[029] Additionally, the parsing module 210, may parse the application code component of the repetitive pattern to identify the plurality of DOM elements. In should be noted that, the parsing technique may be operated using any kind of server client architecture where connection may be established via network protocol such as TCP/IP, UDP, SNMP, ICMP, SSL and HTTP. Thereafter, the parsing module 210 may send the plurality of DOM elements to the state detection module 212 for storing. It should be noted that, the connection between the parsing module 210 and the state detection module 212 may be established by employing connection protocols using standard data connections means such as wireless or wired connections. Examples of the connection protocol may include, but is not limited to Serial Advanced Technology Attachment (SATA), Integrated Drive Electronics (IDE), IEEE-1394, Universal Serial Bus (USB), fiber channel, Small Computer Systems Interface (SCSI), STD Bus, RS-232, RS-422, RS-485, I2C, SPI, Microwire,1-Wire, IEEE 1284, Intel Quick Path Interconnect, InfiniBand, and PCIe.
[030] The state detection module 212 may be configured to receive the plurality of DOM elements parsed from the parsing module 210. The state detection module 212 may also be configured to determine a frequency of change in state of each of the plurality of DOM elements. It should be noted that, the change in state of each of the plurality of DOM elements may be either static state or dynamic state. In an embodiment, a set of static DOM elements from the plurality of DOM elements may be identified directly when the change in state may be static state for each element in the set of static DOM elements. In another embodiment, a set of dynamic DOM elements from the plurality of DOM elements may be identified based on a path when the change in the state may be dynamic state for each element in the set of dynamic DOM elements. Moreover, the static state for the set of static DOM elements may be determined by the parsing module 210 based on a same data types, a same class types, and other similar parameters. Additionally, based on the same data types, a change in data types may be determined for each element in the set of dynamic DOM elements.
[031] In addition, the state detection module 212 may also be configured to generate a pointer corresponding to the application code component (also referred to as thread) for each element in the set of dynamic DOM elements. The pointer corresponding to the application code component may be generated in order to navigate the pointer corresponding to each of the application code component. By way of an example, the pointer may be generated for each such thread, in order to navigate the pointer corresponding to each such thread. Moreover, the state detection module 212 may also be configured to determine the path corresponding to each element in the set of dynamic elements.
[032] Thereafter, the state detection module 212 may be configured to send the frequency of change in state determined for each element in the set of dynamic DOM elements along with the set of dynamic DOM elements to the classifying module 214 for training purpose. It should be noted that, the connection between the state detection module 212 and the classifying module 214 may be established by employing connection protocols via standard data connection means such as wireless or wired connections. Examples of the connection protocol may include, but is not limited to Serial Advanced Technology Attachment (SATA), Integrated Drive Electronics (IDE), IEEE-1394, Universal Serial Bus (USB), fiber channel, Small Computer Systems Interface (SCSI), STD Bus, RS-232, RS-422, RS-485, I2C, SPI, Microwire,1-Wire, IEEE 1284, Intel Quick Path Interconnect, InfiniBand, and PCIe.
[033] The classifying module 214 may be configured to receive the frequency of change in state determined for each element in the set of dynamic DOM elements along with the set of dynamic DOM elements and the associated application code component. Further, the classifying module 214 may be configured to learn rules for classifying each element in the set of dynamic DOM elements into multiple rules. The rules for the particular set of dynamic DOM elements may be determined that may be used for execution post learning the element. The rules may also be defined as an action, that defined by the end user or may be by optimized testing interface developers. The rules learned may also be referred as a plurality of repetitive pattern categories. By way of an example, each element in the set of dynamic DOM elements may be classified into the plurality of repetitive pattern categories based on prediction of similarity between two or more elements. Moreover, the similarity between the two or more elements may be determined based on a similarity threshold. In an embodiment, there may be multiple rules for the same application code component. In another embodiment, there may be multiple rules for the different application code component.
[034] Additionally, the classifying module 214 may also be configured to determine a correct rule corresponding to each element in the set of dynamic DOM elements. The correct rule corresponding to each element may be determined based on selection of element and a corresponding sequence learnt. Further, the classifying module 214 may be configured to generate an output sequence for each element in the set of dynamic DOM elements based on an input sequence. It should be noted that, for 90% of infrastructure automation, the user interface and each element of the set of dynamic DOM elements may be same as previous one. Moreover, when the user interface for the element in the set of dynamic DOM elements is same, than overall output prediction becomes easy. This in turn may help to classify the element with 100% accuracy.
[035] When the user interface corresponding to each element is different, but functionality is similar then, an error rate may exist. Moreover, when the error rate exists, an accuracy for classifying the elements may range from 86% - 90%, for example. The changes in the user interface may happen when a display unit may change, or a dimension of the display unit may change. Hence, in order to fulfill the changes, the classifying module 214 may also learn post few occurrences of the sequence. Based on occurrences learnt, the classification module 214 may understand that the change in the display unit or the dimension has changed the element. Thereafter, the classifying module 214 may be configured to send a predicted misidentification to the error correcting module 216 in order to compare with desired output.
[036] The error correcting module 216 may be configured to receive the predicted misidentification from the classification module 214. The predicted misidentification may be used by the error correcting module 216 to identify error. Further, the error correcting module 216 may be configured to manage span of error by reducing gap between errors based on a predicted sequence and an actual sequence for each element of the set of dynamic DOM element. As will be appreciated, the gap between errors may be reduced largely using backpropagation mechanism. The error-correcting unit may also be configured to proceed with the identified error based on the element mapped from the set of dynamic DOM element. In an embodiment, the backpropagation mechanism may reduce the errors based on backtracking a token corresponding to the element. The correction of error may also include some of real time forward tracking as well as back tracking.
[037] Moreover, some model errors (also referred as formulation error) may also exists, that may be resolved based on user input. It should be noted that, the formulation errors are based on model used in machine learning or deep learning technique. Example of the model that may be used in proposed mechanism may be Long Short-Term Memory (LSTM) model. The LSTM model may have vanishing gradient problem as a bug. Therefore, in order to improve and update the LSTM model, a dataset that includes the set of dynamic DOM elements and the associated application code may be correctly trained by the training module 218. The training module 218 may be developed based on knowledge of determining the different formulation errors in the LSTM model. The formulation errors may be coded based on usability criteria.
[038] Further, the training module 218 may be subjected to a change, in order to understand the formulation error in machine-learning unit of the training module 218. Moreover, the formulation error of other supervised learning may be different. Therefore, a user may need to make suitable changes in the training module 218 based on incoming formulation error. Thereafter, the error correcting module 216 may send back an output corresponding to the error corrected to the classification module 214. In addition, connection between the error correction module 216 and the training module 218 may undergo through direct bus communication. The connection between the error correction module 216 and the training module 218 may also be established via direct bus communication or via network protocols. The network protocols may include, but are not limited to TCP/IP, UDP, SNMP, ICMP, SSL and HTTP
[039] The training module 218 may be configured to receive the set of dynamic DOM elements along with the frequency of change in state for training from the state detection module 212 as an input. The training module 218 may use the LSTM model for learning the set of dynamic DOM elements along with the associated frequency of change of state. The LSTM model may also receive historic data. Further, based on historic data and the frequency of change in state, the LSTM model may learn the sequence pattern of each element in the set of dynamic DOM elements. Moreover, training done by the training module 218 may include different data types as well as different sources and sequence of data. In an embodiment, different data types may correspond to the plurality of repetitive patterns and the data may correspond to the set of dynamic DOM elements. Therefore, the training module 218 may be configured to manage and arrange different data types as well as different sources and sequence of data accordingly. The training of the data is based on past historic sequence collection of the data. The training module may also be configured to segregate the data based on source and difference in the sequence of training data. Further, the classifying module 214 may determine difference based on the sequence learned and trained by the LSTM model.
[040] The optimized testing interface 220 interface may be designed to optimize repetitive task of humans with a standard process and use-case designed by the developers. The optimized testing interface 220 may be used to optimize multiple test cases. The test cases may be used by tester to avoid manual testing. The test cases may include the plurality of repetitive tasks. The optimized testing interface 220 may provide the plurality of DOM elements to the web UI extracting module 204. The plurality of DOM elements may be defined in use case of the optimized testing interface 220. The plurality of DOM elements may be traced, parsed, state detected, and trained. In addition, the plurality of DOM elements may be provided with a trigger. Thereafter, based on the trigger, the optimized testing interface 220 may determine next set of actions as well as next set of elements that may need to be used. It should be noted that, based on above technique of providing the trigger to determine next set of set of actions and elements, the optimized testing interface 220 may enable automatic learning of the use case for testing or developing automated software. The above technique may provide the plurality of DOM elements as an input to the optimized testing interface 220 in order to obtain an output. The output includes the next set of actions as the trigger for the next set of elements it has to operate based upon the next set of actions.
[041] By way of an example, consider a use case to develop a software that may automate a login page, that may be performed without any manual intervention. In such use case, the automated software may provide the domain address and the web-driver details. In addition, sometimes the optimized testing interface 220 may provide details about a plurality of DOM elements associated with the use case. The web UI extracting module 204 may receive the use case as an input and may proceed with the extraction of a plurality of DOM elements. Further, the plurality of DOM elements may be followed by parsing, state detection, training, error correction, and with classification based upon the rules. Thereafter, the plurality of DOM elements may act as the trigger that may be used by the optimized testing interface 220. However, for some use case, the classifying module 214 may be directly called for executing the set of action associated to the webpage (example, the login page) provided, when the rules are already stored.
[042] Referring now to FIG. 3, a flowchart of a method for automating repetitive task on a user interface is illustrated, in accordance with an embodiment. At step 302, the plurality of DOM elements may be identified from the repetitive pattern. The plurality of DOM elements may be identified based on parsing of the application code component of the repetitive pattern. In order to identify the plurality of DOM elements, a dataset that includes the webpage details of the repetitive patterns along with the application code component may be received. It should be noted that the application code component may be a code in any language. Once the dataset is received, the plurality of DOM elements may be traced from the application code component by parsing.
[043] In an embodiment, the application code component may be determined with the HTML and the XML based language. In similar embodiments, the plurality of DOM elements may be determined based on the datatype of each of the plurality of DOM elements. Example of the datatypes for the element may include, but is not limited to, class id, name, tag name, CSS selector, xpath, link text, and partial link text. Moreover, a declaration with these datatypes for the elements may be extracted and stored as dummy variables in the data repository 208. Thereafter, based on the datatypes for the elements stored, the application code component may be parsed to drive the plurality of DOM elements. The plurality of DOM elements may be generated based on a sequence token. The sequence may include combination of the derived element starting from object tracking until the end of the element. However, sometimes a change may exist within the sequence, and that may be traced based on occurrence of the sequence. Moreover, the dataset may be reformed to form another sequence before performing training, based on a predefined threshold point for the occurrence of the sequence. Based on reformation of the dataset, diversified element within same sequence may be traced thereby enhancing the automation technique. In addition, post creating the sequence a heap may be created that may be followed for training and executing each of the plurality of DOM element in the same heap.
[044] With reference to FIG. 2, the element tracking module 206 may receive the dataset from the web UI extraction module 204. Thereafter, the frequency of change in state may be determined for each of the plurality of DOM elements. In conjunction to FIG. 2, the frequency of change in state may be tracked by the element tracking module 206. It should be noted that, the frequency of change in state corresponding to each element in the plurality of DOM elements may be either one of the static state or the dynamic state. Thereafter, based on the frequency of change in state corresponding to the plurality of DOM elements, the LSTM model may be trained. It should be noted that, a state of each element in plurality of DOM element is very important to understand the element validity and verification. Further, the state is determined based on the change in the plurality of DOM elements. By way of an example, if an element value corresponding to the plurality of DOM elements, changes every time on execution of the application code component in historic data then the element may be referred as dynamic. Similarly, if the element value corresponding to the plurality of DOM elements remains same as in historic data, then element may be referred as static. The static state may be directly consumed based on the element.
[045] At step 304, the set of dynamic DOM elements may be identified from the plurality of DOM elements. The set of dynamic DOM elements may be identified based on the frequency of change in state determined. The frequency of change in state for each element of the set of dynamic DOM elements may be above a first pre-defined threshold. Moreover, execution of the set of dynamic DOM elements may be based on the sequence within the repetitive pattern. At step 306, the path and the path position may be determined within the application code component of the repetitive pattern for each of the set of dynamic DOM elements. With reference to FIG. 2, the path position may correspond to the location. In an embodiment, the path may be determined based on the pointer and the path position may be determined by a tree navigating to that pointer. The pointer may be included in the element and other thread. Based on the pointer a passive path that may locate the respective pointer may be determined. Further, based on the determined passive path the tree navigation may be initiated from that pointer.
[046] By way of an example, suppose the application code component may be in HTML language. The DIV element may be considered as a main component of the application code component. In addition, another component based on the DIV element, may include, but is not limited to, header, sub div or any other component. Moreover, based on the pointer, a first DIV element may be captured. The first DIV element captured along with the datatype may be fed to parsing module 210 to check whether the DOM element corresponding to the first DIV may be detectable or not. If the DOM element may not be detectable then the tree may navigate to next DIV element. A cycle of detecting the DOM element may continue until the DOM element may be detectable from the application code associated to the user interface. Sometimes, after many cycles, the DOM elements may still not be detectable because of Iframe. It should be noted that, the Iframe is another core component of the HTML. As in case of Iframe, while moving to next DIV an iframe element needs to be detected first. Thereafter, based on the Iframe element the next DIV need to be called out.
[047] As will be appreciated, apart from the HTML being most widely used in web designing, a system-based application may also be easily detected using standard protocols. Moreover, for every action, it may be recommended to define a function as per coding standard. In addition, in the system-based application there may be no DIV but the control of actions may be detected based on the function and the coding standards. Example for the application code component in the HTML is represented below:
Find-web-element (by-iframe (by-div (by-class= “XXXX”)))
{

Find-web-element (by-div (by-id = “XXXX”))

[048] Similarly, for the system-based application, the application code component may look similar, only coding features and associated parameters may change. In addition, the passive path of sequence is generated for each element of the set of dynamic elements. Thereafter, the set of dynamic elements are stored in the data repository 208 for training purpose. The sequence of each element may happen as being detected by action of the end user. Moreover, the sequence of each element may be arranged based on the occurrence of element.
[049] At step 308, an Artificial Intelligence (AI) model may be trained for each of the repetitive pattern. The training of the AI model may be done for each element in the set of dynamic DOM element based on the path and the path position within the application code component. In addition, the sequence corresponding to each element in the set of dynamic DOM elements may be also used training. By way of an example, the LSTM model may be trained based on the sequence, the path and the path position for each element in the set of dynamic DOM elements. Moreover, in order to learn the long chain of the sequence, deep neural network may be used to identify and remember the sequence. Further, the identified sequences may be grouped together to form a new dataset. The new dataset may be fed into a first layer or input layer of neurons of the LSTM model. The neurons may include the input layer with three gates. A first gate may correspond to an update gate. The first gate may receive the new dataset as input that may be further fed into the neurons of the input layer. The first gate may then measure volume based the new dataset received. Thereafter, based on the volume determined, weights and initial neurons may be calculated for the new dataset. Further, a result determined by the input layer may be forwarded to a forget layer of the neurons of the LSTM model.
[050] The forget layer may then determine relevancy of the identified sequence corresponding to the set of dynamic DOM elements. Further, the forget layer may remove all unwanted chunks of data and errors from the result of input layer. The unwanted chunks may need to be removed, in order to ensure the high level of accuracy of the LSTM model. Moreover, post refining the result received from the input layer, by the forget layer, the result may be added to an output layer. The output layer may predict the next set of action based on timestamps of past historic timestamp input. The output layer may act as an encoder and a decoder, where the identified sequence corresponding to the set of dynamic DOM elements may be provided as an input and the next set of actions may be generated as an output. The output may be generated based on the length of the sequence. It should be noted that, other than LSTM model many other models may be used, but the LSTM model provides the best accuracy for the proposed mechanism. Moreover, the LSTM model may transform itself into one-one distribution when a single sequence may be provided as input. Similarly, when more than one sequence is provided as input than the output generated may be same in number as the input sequence. At step 310, a code for each of the set of dynamic DOM elements may be automatically executed based on the associated path and the path position. Moreover, the code for each of the set of dynamic DOM elements may be executed based on conformance of the sequence within the repetitive pattern.
[051] Referring now to FIG.4, a flowchart of a method for performing incremental learning of an AI model based on training of the AI model for a plurality of repetitive tasks (also referred as repetitive patterns) is illustrated, in accordance with an embodiment. At step 402, the plurality of repetitive patterns may be identified within the user interface. The repetitive patterns may be identified when the frequency of repetition for each of the repetitive pattern may be above a second predefined threshold. Moreover, the plurality of repetitive patterns may be identified from a set of data that includes details of an application code associated with the user interface. Once the repetitive patterns are identified then, at step 404, the similarity between the set of dynamic DOM elements may be identified within each of the plurality of repetitive patterns. At step 406, each of the plurality of repetitive patterns may be classified into the plurality of patterns categories based on the determined similarity. The classification may be done by categorizing two or more repetitive patterns in a pattern category from the plurality of pattern categories. The two or more repetitive patterns may be classified in the pattern category, when the similarity between the two or more repetitive patterns is above the similarity threshold. At step 408, the AI model may be trained to uniquely identify the pattern category associated with each of the plurality of repetitive patterns. Once the AI model is trained, at step 410, the misidentification of the pattern category associated with the repetitive pattern may be determined by the AI model. Thereafter, at step 412, incremental learning of the AI model may be performed based on the determined misidentification.
[052] Referring now to FIG. 5, a flowchart of a method for determining a set of dynamic DOM elements from a plurality of DOM elements is illustrated, in accordance with an embodiment. At step 502, the plurality of DOM elements may be identified from the repetitive pattern. At step 504, an element value associated with an attribute may be determined from each of the plurality of DOM elements. On determining the element value and the associated attribute value, at step 506, the frequency of change of the element value corresponding to the attribute may be determined for each of the plurality of DOM elements. The frequency of change of the element value may be determined based on a comparison of each of the element value with corresponding historic element values. The change in an element value associated with an attribute of a DOM element may correspond to change of state of the DOM element. At step 508, we may assume that the total number of plurality of DOM elements may be ‘N’. In addition, ‘n’ may be the ‘nth’ DOM element from the plurality of DOM elements. Thereafter, at step 510, the DOM element ‘n’ may be initialized to one i.e., n =1. It should be noted that, the value initialized to ‘n’ may keep on increasing once the change of state is determined for n-1th DOM element.
[053] At step 512, the frequency of change of state may be evaluated for the nth DOM element based on the first predefined threshold. In an embodiment, when the frequency of change of state of the nth DOM element is evaluated to be greater than the first predefined threshold, then at step 514, the nth DOM element may be identified as a dynamic DOM element. However, the frequency of change of the element value with the associated attribute value for the nth DOM element may be evaluated based on the first predefined threshold. Moreover, if the frequency of change of the element value of the nth DOM element is determined to be greater than the first predefined threshold, then the nth DOM element may be listed as the dynamic DOM element. At step 516, the identified nth DOM element (listed as the dynamic DOM element) may be added to the set of dynamic DOM elements.
[054] In another embodiment, when the frequency of change of state of the nth DOM element is evaluated to be less than or equal to the first predefined threshold, then at step 518, the nth DOM element may be identified as a static DOM element. However, the frequency of change of the element value with the associated attribute value for the nth DOM element may be evaluated based on the first predefined threshold. Moreover, if the frequency of change of the element value of the nth DOM element is determined to be less than or equal to the first predefined threshold, then the nth DOM element may be listed as the static DOM element. At step 520, the identified nth DOM element (listed as the static DOM element) may be added to the set of static DOM elements. At step 522, the value one initialized to ‘n’ may be incremented by one represented as ‘n = n+1’.
[055] Referring now to FIG. 6, a flowchart of a method for improving an accuracy of an AI model based on an error identified is illustrated, in accordance with an embodiment. At step 602, the code for each of the set of dynamic DOM elements may be automatically executed based on the associated path and the path position. Moreover, the code for each of the set of dynamic DOM elements may be executed based on conformance of the sequence within the repetitive pattern. Once the automatic execution of the code starts, then at step 604, error may be identified in automatic execution of the code for each of the set of dynamic DOM elements. The error may be identified by the AI model in conformance with the sequence within the repetitive pattern. It should be noted that the error may be traced by the error correcting module 216. The error may be captured based on the actual sequence and the predicted sequence for each of the set of dynamic DOM elements. By way of an example, the error may be captured by determining a variance in the actual sequence and the predicted sequence. The variance determined may be a binary number or a probabilistic number based on different coding standards and convention.
[056] It should be noted that, the error tracing may be generally performed based on the back-propagation mechanism or by using similar other mechanism like regression analysis. Moreover, the back-propagation mechanism may be generally considered as in the proposed mechanism, the deep neural network may be used. The back-propagation mechanism may be performed during testing phase of the AI model in order to achieve 100% accuracy of the AI model. Therefore, the variance in the predicted sequence and the actual sequence may be noticed before releasing the AI model to operate in operation.
[057] At step 606, incremental learning for the AI model may be performed based on the error identified, in order to improve the accuracy of the AI model. In addition, the accuracy of the AI model may be improved to execute the code for each of the set of dynamic DOM elements in conformance with the sequence within the repetitive pattern. Moreover, the accuracy may be improved by replacing the error with the variance determined to automatically generate correct output. This may ensure consecutive training of the AI model. Once the error is eliminated, then the AI model may be used for real time application.
[058] An applicability of proposed invention in the present disclosure is explained via an example. Consider a scenario, where the repetitive task (also referred as repetitive pattern) that may need to be automated may correspond to a login process for a particular web page. To automate the login process, the plurality of DOM elements may be traced from a login web page. The plurality of DOM elements for the login web page may be determined by parsing the application code component corresponding to the login process. Further, an element value with an associated attribute value may be determined from the application code component for the login web page. The element value here may correspond to a password and the associated attribute may correspond to an input to username. The password and the username may be predefined by the end user.
[059] Alternatively, the password and the username may be learned by the AI model based on pattern detection. Thereafter, the AI model may fetch a code that may control process of login button. Once the code is determined, a predicted action for automating the process of login may include clicking on the login button. This action may be performed by the classification module 214. Alternatively, the classification module 214 may generate the trigger to the optimized testing interface 220. In other words, the classification module 214 may pass the control to the optimized testing interface 220. This cycle may keep on going.
[060] Various embodiments of the invention provide method and system for automating repetitive task on a user interface. The method and system may first identify a plurality of DOM elements from the repetitive patterns. The plurality of DOM elements may be identified by parsing an application code component of the repetitive pattern. The method and system may then identify a set of dynamic DOM elements from the plurality of DOM elements. Further, the method may identify a path and a path position within the application code component of the repetitive pattern for each of the set of repetitive patterns. Thereafter, the method and system may perform training of an Artificial Intelligence (AI) model to identify the repetitive pattern, and the corresponding path and the path position within the application code component for each of the set of dynamic DOM elements. Additionally, the method and the system may execute a code automatically for each of the set of dynamic DOM elements based on the associated path and path position.
[061] The benefit of the invention is that, the present invention may automate RPA process and tracking of the application code component (also referred as web element code) without human intervention. The present invention may also reduce manual efforts and time taken to develop an RPA bot. Moreover, the present invention may support auto code generation technique, thereby eliminating the need to fetch and develop the RPA bots. Additionally, the present invention explores and develop multiple assistance model that may be applicable for system-based application as well as part from web applications. Moreover, the present invention supports timely proportional changes corresponding to the change in web-element (i.e., the DOM element) of the user interface. Furthermore, the present invention disclosed may reduce manual efforts of the end user for performing the repetitive tasks on daily basis.
[062] The specification has described method and system for automating a repetitive task on a user interface. The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments.
[063] It is intended that the disclosure and examples be considered as exemplary only, with a true scope and spirit of disclosed embodiments being indicated by the following claims.