In the rapidly evolving digital era, industries increasingly harness technology to optimize operations. The banking sector, in particular, stands out as a prominent example, having integrated Artificial Intelligence (AI) to streamline processes and improve efficiency. Our study focuses on one key aspect: automating loan management, specifically through Intelligent Document Processing (IDP). While automation technologies have been widely studied, a notable gap exists in sector-specific knowledge, especially within the banking industry. This paper conducts a Systematic Literature Review (SLR), examining 48 primary studies, to analyze the state-of-the-art of this problem. This comprehensive analysis reveals how IDP reshapes banking processes, providing sector-specific insights. Our findings reveal the profound impact of automation in banking, along with 8 notable challenges that remain to be addressed. This contributes to future research and enriches our understanding of IDP’s current and potential applications in the sector.

Robotic Process Automation (RPA) enables subject matter experts to use the graphical user interface as a means to automate and integrate systems. This is a fast method to automate repetitive, mundane tasks. To avoid constructing a software robot from scratch, Task Mining approaches can be used to monitor human behavior through a series of timestamped events, such as mouse clicks and keystrokes. From a so-called User Interface log (UI Log), it is possible to automatically discover the process model behind this behavior. However, when the discovered process model shows different process variants, it is hard to determine what drives a human’s decision to execute one variant over the other. Existing approaches do analyze the UI Log in search for the underlying rules, but neglect what can be seen on the screen. As a result, a major part of the human decision-making remains hidden. To address this gap, this paper describes a Task Mining framework that uses the screenshot of each event in the UI Log as an additional source of information. From such an enriched UI Log, by using image-processing techniques and Machine Learning algorithms, a decision tree is created, which offers a more complete explanation of the human decision-making process. The presented framework can express the decision tree graphically, explicitly identifying which elements in the screenshots are relevant to make the decision. The framework has been evaluated through a case study that involves a process with real-life screenshots. The results indicate a satisfactorily high accuracy of the overall approach, even if a small UI Log is used. The evaluation also identifies challenges for applying the framework in a real-life setting when a high density of interface elements is present.

Abstract The robotic process automation (RPA) paradigm is a discipline that is becoming increasingly popular thanks to the great interest shown by the industry. In such context, RPA solutions based on artificial intelligence, i.e., cognitive solutions, are receiving increasing attention. In a cognitive RPA project, the RPA developer is in charge of selecting the most suitable components that solve specific tasks from the sets of components provided by different RPA platforms. This selection is very challenging, especially since there is no homogeneity in component names or component classifications. Such a situation turns an RPA project’s development into a time-consuming, error-prone, and very tedious process. Therefore, supporting the RPA developer in developing a cognitive RPA project is desired. The industry has also pointed out this need. This work presents a proposal for supporting the users in developing a cognitive RPA project. To be more precise, an incremental method to automatically generate taxonomies from cognitive RPA platforms is proposed. Such taxonomies can be dynamically adapted when necessary. In previous work, the initial aspects of this research were presented. However, the current work greatly enhances such previous work by: (1) extending the proposed method to improve the management of real-world use cases from industry, (2) developing a proof-of-concept tool that is based on the proposed approach, (3) validating the proposed method by applying it to real-world use cases from industry, and (4) performing a literature review on related topics. The results obtained are auspicious and demonstrate that the proposed approach substantially improves the support given to users during the development of a cognitive RPA project.

Business process analysis is a key factor in the lifecycle of Robotic Process Automation. Currently, task mining techniques provide mechanisms to analyze information about the process tasks to be automated, e.g., identify repetitive tasks or process variations. Existing proposals mainly rely on the user interactions with the UIs of the system (i.e., keyboard and mouse level) and information that can be gathered from them (e.g., the window name) which is stored in a UI event log. In some contexts, the latter information is limited because the system is accessed through virtualized environments (e.g., Citrix or Teamviewer). Other approaches extend the UI Log, including screenshots to address this issue. Regardless of the context, the aim is to store as much information as possible in the UI Log so that is can be analyzed later on, e.g., by extracting features from the screenshots. This amount of information can introduce much noise in the log that messes up what is relevant to the process. To amend this, the current approach proposes a method to include a gaze analyzer, which helps to identify which is process-relevant information between all the information. More precisely, the proposal extends the UI Log definition with the attention change level, which records when the user’s attention changes from one element on the screen to another. This paper sets the research settings for the approach and enumerates the future steps to conduct it.

Nowadays, Machine Learning (ML) algorithms are being widely applied in virtually all possible scenarios. However, developing a ML project entails the effort of many ML experts who have to select and configure the appropriate algorithm to process the data to learn from, between other things. Since there exist thousands of algorithms, it becomes a time-consuming and challenging task. To this end, recently, AutoML emerged to provide mechanisms to automate parts of this process. However, most of the efforts focus on applying brute force procedures to try different algorithms or configuration and select the one which gives better results. To make a smarter and more efficient selection, a repository of knowledge is necessary. To this end, this paper proposes (1) an approach towards a common language to consolidate the current distributed knowledge sources related the algorithm selection in ML, and (2) a method to join the knowledge gathered through this language in a unified store that can be exploited later on, and (3) a traceability links maintenance. The preliminary evaluations of this approach allow to create a unified store collecting the knowledge of 13 different sources and to identify a bunch of research lines to conduct.

Nowadays, Machine Learning (ML) algorithms are being widely applied in virtually all possible scenarios. \nHowever, developing a ML project entails the effort of many ML experts who have to select and configure \nthe appropriate algorithm to process the data to learn from, between other things. Since there exist thousands \nof algorithms, it becomes a time-consuming and challenging task. To this end, recently, AutoML emerged to \nprovide mechanisms to automate parts of this process. However, most of the efforts focus on applying brute \nforce procedures to try different algorithms or configuration and select the one which gives better results. \nTo make a smarter and more efficient selection, a repository of knowledge is necessary. To this end, this \npaper proposes (1) an approach towards a common language to consolidate the current distributed knowledge \nsources related the algorithm selection in ML, and (2) a method to join the knowledge gathered through this \nlanguage in a unified store that can be exploited later on. The preliminary evaluations of this approach allow \nto create a unified store collecting the knowledge of 13 different sources and to identify a bunch of research \nlines to conduct.